The 8 best AI development companies in Australia

An AI development company is a firm that designs, writes and ships custom AI software for a client, such as agents, models and the integrations that connect them to the client's systems, rather than one that only advises or resells a product.

On a 60-point test of engineering, published proof, published prices, ownership and team, EB Pearls scores highest among Australian AI development companies at 58. 4mation and Advancer tie on 55, and 4mation takes second place on the evidence tie-break. Hourglass AI and Synap AI tie on 53, level on both tie-breaks, and are listed fourth and fifth alphabetically.

The best AI development companies in Australia, ranked

How each one scores

Points by test, scored 1 October 2026

Points for 8 companies on 5 tests, with the total out of 60.

EB Pearls	4mation	Advancer	Hourglass AI	Synap AI	Team 400	AI Advancements	V2 AI
Engineering20 points
20 of 20	20 of 20	20 of 20	20 of 20	20 of 20	20 of 20	20 of 20	20 of 20
Evidence10 points
10 of 10	8 of 10	5 of 10	8 of 10	8 of 10	7 of 10	10 of 10	5 of 10
Published pricing10 points
10 of 10	7 of 10	10 of 10	10 of 10	10 of 10	7 of 10	4 of 10	0 of 10
Ownership10 points
8 of 10	10 of 10	10 of 10	10 of 10	10 of 10	6 of 10	4 of 10	10 of 10
Team and track record10 points
10 of 10	10 of 10	10 of 10	5 of 10	5 of 10	8 of 10	8 of 10	8 of 10
Total scoreout of 60
58 of 601st	55 of 602nd	55 of 603rd	53 of 604th	53 of 605th	48 of 606th	46 of 607th	43 of 608th

1. EB Pearls

Rank 1

EB Pearls

Best for

Funded companies and enterprises with a build budget from about $50,000 that want a large, long-running engineering team on a fixed-scope build

Location

Suite 1A7, 410 Elizabeth Street, Surry Hills, Sydney, with engineering in Kathmandu

Engineering20/20

Evidence10/10

Published pricing10/10

Ownership8/10

Team and track record10/10

Total score58/60

EB Pearls has been Australian owned and operated since 2004. Its 320-plus engineers are full-time staff in Kathmandu, Nepal, and the account manager and engagement lead sit in Sydney. Its AI practice prices each paid stage: a working prototype from $5,000, scoping from $3,000, a build from $50,000 and ongoing model care from $800. On the same page, EML reports a 65% cut in claims resolution time, and AWN reports 94% accuracy and 60% faster wool appraisals from a valuation model trained on auction data. All IP is assigned to the client at the end of every milestone, and the Surry Hills office holds a Google rating of 4.6 on 23 reviews.

• Beats the field on: evidence, 10 of 10, level with AI Advancements, and published prices, 10 of 10, level with Advancer, Synap AI and Hourglass AI.
• Loses on: ownership, 8. None of the pages read says the system runs on the client's own cloud accounts.

Free 30-minute call

Ask Hourglass AI what it would build first, and what it would cost

Bring one job your team repeats every week and the systems it runs through, such as your CRM or accounting software. In 30 minutes we check whether an AI build pays back, what it would connect to and how its cost compares with the prices on this page. If the numbers don't work, we'll say so.

For Australian companies of 10 to 100 people. We reply the same day, and if we're a fit, your AI Audit can start the same week.

Book a 30-minute call

2. 4mation

Rank 2

4mation

Best for

Mid-sized organisations that want fixed-price AI builds from a Sydney team, with the IP assigned to them at the end

Location

Level 17, 25 Bligh Street, Sydney

Engineering20/20

Evidence8/10

Published pricing7/10

Ownership10/10

Team and track record10/10

Total score55/60

4mation publishes its prices. AI consulting packages run $6,000 to $12,000 plus GST, chatbot builds are $20,000 and $40,000, and an end-to-end custom agent workflow is $25,000 plus GST. For the ABC it automated planned outage approvals, cutting manual handling time by 60% and automating 75% of approvals. The QV IQ chatbot it built for Ego Pharmaceuticals handled 684 chats at 97% accuracy in its first month. Its guide to software IP says ownership is fully assigned to the client when a project ends, and the Sydney firm has more than 40 staff.

• Beats the field on: ownership, 10 of 10, one of 11 companies on full marks, and team, 10 of 10, one of eight.
• Loses on: published prices, 7. Its Service Desk support tiers are named but carry no price.

3. Advancer

Rank 3

Advancer

Best for

Queensland businesses that want a voice or process agent built, documented and handed to their own team to run

Location

Fortitude Valley, Brisbane

Engineering20/20

Evidence5/10

Published pricing10/10

Ownership10/10

Team and track record10/10

Total score55/60

Advancer was founded in Brisbane in 2015. An AI Readiness Audit starts from $2,500. Simple automation projects run $5,000 to $15,000, and multi-system AI agents typically run $20,000 to $80,000 or more, deployed on-premise or in the client's own cloud if required. Its Voice AI agents for Bartons Motor Group handle thousands of service calls. The two Voice AI agents it built for The Tivoli and The Princess Theatre are now maintained by the Tivoli team in-house. Its platform page offers to document and train for client ownership from day one.

• Beats the field on: published prices, 10 of 10, one of four companies on full marks, plus ownership and team, 10 of 10 each.
• Loses on: evidence, 5. Bartons Motor Group is its one named client with a figure. Its larger savings figures are anonymised.

4. Hourglass AI

Rank 4

Hourglass AI

Best for

Australian companies of 10 to 100 people that want fixed-fee AI builds and automations running on their own servers

Location

Sydney and Melbourne

Engineering20/20

Evidence8/10

Published pricing10/10

Ownership10/10

Team and track record5/10

Total score53/60

Hourglass AI builds for Australian companies of 10 to 100 people. Builds are published at $5,000 to $25,000 each and go live in one to two weeks, inside the tools a team already uses, with no migration. Maintenance is about $1,000 a month per build, and the audit that comes first is a flat fee.

Everything it builds runs on the client's own servers and accounts, and the client owns all of the IP. The Company AI Brain at the centre is plain text files on the client's server, readable by Claude, GPT, Gemini or a local model, so no build is locked into one AI model.

Its builds take admin out of growing teams. Integr8 cut its admin costs by 40%, and Voyagin Australia cut its admin time by 75%, both in the clients' own words. Startup Daily reported in April 2026 that Schweigen, a multi-billion-dollar appliance retailer, runs an AI customer support system Hourglass built, live on its own site. At one client, an executive assistant's 500-plus emails a day are now run by its AI Inbox, which spread from 5 users to a 35-person office in 7 weeks. Another client went from 2 people using AI to 28, with 1,360 skill calls in 30 days.

Each engagement starts with an audit that ends in an interactive client portal rather than a PDF report, and about 85% of the businesses that take a full audit go on to a build. Co-founder Michael Batko came to Hourglass AI after eight years leading Startmate, which backed 228 companies with first cheques in that time. He told AI Down Under in August 2026 that about 95% of Hourglass AI's clients arrive by referral. Co-founder Finlay Ekins put its client count at 35 companies in June 2026.

• Strongest on: engineering, published prices and ownership, with full marks on all three, 40 of 40 points, and evidence, 8 of 10, level with 4mation, Synap AI and Mantel Group.

5. Synap AI

Rank 5

Synap AI

Best for

Businesses of 5 to 200 staff that want a fixed-price build from a founder-led studio

Location

Level 8, 220 Collins Street, Melbourne

Engineering20/20

Evidence8/10

Published pricing10/10

Ownership10/10

Team and track record5/10

Total score53/60

Synap AI publishes every price on its home page. An AI Readiness Assessment is $950, custom software runs from $1,200 to $50,000 or more, and Fractional AI Advisor retainers run from $1,200 to $15,500 a month. For Dragonfly Remedial, an engineering remediation firm, it built a multimodal report workflow that takes about 30 hours of manual work out of every report. Code, data and hosting sit in the client's own environment. Founder and lead engineer Hamish Mackellar has 25 years of senior engineering experience, and he architects every custom software build himself.

• Beats the field on: published prices and ownership, 10 of 10 each.
• Loses on: team, 5. It has operated since 2025, and its site names no one but its founder.

6. Team 400

Rank 6

Team 400

Best for

Mid-market organisations on Microsoft Azure that want agents and models built by senior engineers

Location

383 George Street, Sydney

Engineering20/20

Evidence7/10

Published pricing7/10

Ownership6/10

Team and track record8/10

Total score48/60

Team 400 built Coast Smoke Alarms a custom field service platform with AI scheduling that automates 80% of its scheduling. Custom AI projects typically cost $50,000 to $250,000, and Fast Packages start from $25,000. It works with Azure AI Foundry, OpenAI, Anthropic Claude and LangChain, and its AI agent FAQ says client data stays within the client's infrastructure. It has been building mobile apps for Australian brands such as Medibank, Kmart and Telstra for more than 15 years, and its Sydney office holds a Google rating of 5.0 on 10 reviews.

• Strongest on: engineering, 20 of 20, full marks like the rest of the top eight.
• Loses on: ownership, 6. None of the pages read says the client owns the code or IP.

7. AI Advancements

Rank 7

AI Advancements

Best for

Western Australian and regulated organisations that want an engineering team embedded alongside their own

Location

Unit 2, 125 Main Street, Osborne Park, Perth

Engineering20/20

Evidence10/10

Published pricing4/10

Ownership4/10

Team and track record8/10

Total score46/60

AI Advancements embedded with Fortescue's fleet control team, where one automation saves about three hours a day at a near-zero error rate. Amplitude Energy's finance team gets a week back every month on balance sheet reconciliation. Every development project begins with a $2,000 paid scoping engagement, credited towards the build, and is deployed within the client's Australian-based infrastructure. The Perth firm has grown alongside its clients for two and a half years and holds a Google rating of 5.0 on 37 reviews.

• Beats the field on: evidence, 10 of 10, level with EB Pearls.
• Loses on: published prices, 4. No build price appears on the pages read.

8. V2 AI

Rank 8

V2 AI

Best for

Enterprises and regulated organisations on Azure, AWS or Databricks that want a production AI system their own team can run after handover

Location

Level 2, 555 Collins Street, Melbourne, with offices in Sydney, Brisbane and Perth

Engineering20/20

Evidence5/10

Published pricing0/10

Ownership10/10

Team and track record8/10

Total score43/60

V2 AI designed and delivered Ada, an adviser digital assistant, for Allianz Retire+. Ada is a production AI capability that gives financial advisers self-service access to product knowledge. For Onyx Gaming it built a unified data platform that set up the retirement of the old reporting stack, with estimated 30% annual savings. Onyx's engineers were involved early, so the client took complete ownership of the platform, with no long-term reliance on V2 AI. At Swinburne University, the team now manages its Databricks environment and predictive models without external support. Consultancy.com.au describes it as a 2023-founded consultancy, and it publishes no prices.

• Beats the field on: ownership, 10 of 10, one of 11 companies on full marks.
• Loses on: published prices, 0. No price appears on any page read.

Also scored

• Also read and scored, below 34: PhoenixDX 33, JPLoft 33, Saigon Technology 32, Jaarvis Technologies 30, Appther 29, Agentive 28, Cognify Digital 28, ZenityAI 28, Vrinsoft 28, Blockstars 26, C9 24, Concetto Labs 23, Supportsoft Technologies 22, Ziff Digital 22, Digitalon AI 22, Avenue Group Australia 22, Apptechies 21, ISH Technologies 18, Source Central 16 and Appello 15.

How they compare

The grid below checks each company's own pages against five questions a buyer asks on the first call: does it publish build prices, do you own the code, does the system run on your own servers, does it offer support after launch, and does it name the people who build? The cost table under it shows what each one charges. All eight name at least one builder, and seven of the eight offer ongoing support on the pages read. Six of the eight publish a build price: AI Advancements quotes its builds after a paid scoping engagement, and V2 AI publishes no prices.

8 companies, side by side

Checked 1 October 2026

QuestionEB Pearls4mationAdvancerHourglass AISynap AITeam 400AI AdvancementsV2 AI

Publishes build prices?A price or range for a build, on its own siteEB Pearls: Yes.4mation: Yes.Advancer: Yes.Hourglass AI: Yes.Synap AI: Yes.Team 400: Yes.AI Advancements: No.V2 AI: No.

Publishes build prices? Each company's answer, from its own pages.

Company	Answer
EB Pearls	Yes. Most builds land between AU$80K and AU$1M+, and its AI page lists paid scoping from $3K and builds from $50K.
4mation	Yes. Publishes AI agent packages from $25,000 and chatbot builds at $20,000 and $40,000.
Advancer	Yes. Publishes $5,000 to $15,000 for simple automations and $20,000 to $80,000+ for multi-system agents.
Hourglass AI	Yes. Builds are $5,000 to $25,000 each, confirmed after a scoping call.
Synap AI	Yes. Publishes custom software from $1,200 to $50,000+.
Team 400	Yes. Publishes $50,000 to $250,000 for custom AI projects, with packages from $25,000.
AI Advancements	No. Publishes only its $2,000 scoping engagement, and builds are quoted after scoping.
V2 AI	No. Publishes no price on the pages read.

Client owns the code and IP?Its own site says the client owns what is builtEB Pearls: Yes.4mation: Yes.Advancer: Yes.Hourglass AI: Yes.Synap AI: Yes.Team 400: Not stated.AI Advancements: Not stated.V2 AI: Yes.

Client owns the code and IP? Each company's answer, from its own pages.

Company	Answer
EB Pearls	Yes. Assigns all IP to the client at the end of every milestone.
4mation	Yes. Its IP guide says ownership is fully assigned to clients on completion.
Advancer	Yes. Will hand ownership to the client if asked at the start.
Hourglass AI	Yes. Says you own all of the IP, and its terms assign custom deliverables on payment.
Synap AI	Yes. Builds live in version control the client can own.
Team 400	Not stated. Not stated on the pages read.
AI Advancements	Not stated. Not stated on the pages read.
V2 AI	Yes. Its Onyx Gaming case study says the client owns the platform, with no long-term reliance on V2 AI.

Runs on your own servers?Its site says the system runs on the client's own servers or accountsEB Pearls: Not stated.4mation: Yes.Advancer: Yes.Hourglass AI: Yes.Synap AI: Yes.Team 400: Yes.AI Advancements: Yes.V2 AI: Yes.

Runs on your own servers? Each company's answer, from its own pages.

Company	Answer
EB Pearls	Not stated. Not stated on the pages read.
4mation	Yes. Its IP guide tells clients to hold the cloud tenancy in their own name.
Advancer	Yes. Can deploy on-premise or in the client's own cloud.
Hourglass AI	Yes. Everything runs on the client's own servers and accounts.
Synap AI	Yes. Code, data and hosting sit in the client's own environment.
Team 400	Yes. Its AI agent FAQ says client data stays within the client's infrastructure.
AI Advancements	Yes. Deploys within the client's Australian-based infrastructure.
V2 AI	Yes. A public sector platform it built runs inside the client's existing Azure, on the client's own billing.

Ongoing support offered?Ongoing support after launch, offered on its own siteEB Pearls: Yes.4mation: Yes.Advancer: Yes.Hourglass AI: Yes.Synap AI: Yes.Team 400: Yes.AI Advancements: Yes.V2 AI: Not stated.

Ongoing support offered? Each company's answer, from its own pages.

Company	Answer
EB Pearls	Yes. Retainer packages after a 30-day post-launch window, from $400.
4mation	Yes. Service Desk support tiers, from incident response to proactive monitoring.
Advancer	Yes. Monitors, supports and iterates on the agents it builds.
Hourglass AI	Yes. Maintenance at about $1,000 a month per build: fixes, model upgrades, improvements.
Synap AI	Yes. Retainers include ongoing execution.
Team 400	Yes. Managed services: monitoring, optimisation, retraining.
AI Advancements	Yes. Embeds a specialist on a time-and-materials basis.
V2 AI	Not stated. Not stated on the pages read.

Named AI builders on staff?Names at least one person in an AI or engineering build roleEB Pearls: Yes.4mation: Yes.Advancer: Yes.Hourglass AI: Yes.Synap AI: Yes.Team 400: Yes.AI Advancements: Yes.V2 AI: Yes.

Named AI builders on staff? Each company's answer, from its own pages.

Company	Answer
EB Pearls	Yes. Names its senior engineering managers, with engineering based in Kathmandu.
4mation	Yes. Names its Head of Engineering.
Advancer	Yes. Names AI Engineer Rafael Golin.
Hourglass AI	Yes. Names Jeremy Sharpe, AI Builder, and co-founder Finlay Ekins.
Synap AI	Yes. Founder Hamish Mackellar is the lead engineer.
Team 400	Yes. Founder and CTO Michael Ridland.
AI Advancements	Yes. Names a full-stack engineer specialising in LLMs.
V2 AI	Yes. Names its Delivery Lead on its about page.

Yes answersOut of 5 questions4 yes for EB Pearls5 yes for 4mation5 yes for Advancer5 yes for Hourglass AI5 yes for Synap AI4 yes for Team 4003 yes for AI Advancements3 yes for V2 AI

What an AI development build costs

Published prices, AUD

What an AI development build costs, by company: Entry or audit, Typical build, Ongoing.

Company	Entry or audit	Typical build	Ongoing
EB Pearls	From $3KScoping and agreement on its AI page, and a working prototype is from $5K	AU$80K to AU$1M+A focused MVP to a full enterprise replatform	From $800Evolve phase on its AI page: monitoring, drift checks and retraining
4mation	$6,000 + GSTAI consulting package, rising to $12,000	From $20,000 ex GSTChatbots $20,000 and $40,000, and agents from $25,000	Not published
Advancer	From $2,500AI Readiness Audit	$5,000 to $80,000+Simple automations to multi-system agents	From $8,000 a monthFractional AI Director retainer (advisory)
Hourglass AI	Flat feeAI Audit, credited in full to a build started within 30 days	$5,000 to $25,000Each build, quoted as a range and confirmed after a scoping call	About $1,000 a monthPer build, with AI usage billed to the client's own account
Synap AI	$950AI Readiness Assessment	$1,200 to $50,000+Custom software, typically 2 to 6 weeks	$1,200 to $15,500 a monthFractional AI Advisor retainers, plus GST
Team 400	From $25,000Fast Packages for focused AI solutions	$50,000 to $250,000Custom AI projects	Not published
AI Advancements	$2,000Paid scoping, credited to the build	Not published	Time and materialsEmbedded specialist, no rate published
V2 AI	Not published	Not published	Not published

Which AI development company to pick for your build

• Pick a global consultancy if the job is an enterprise platform across many countries. Those firms are outside the segment this list scores.
• Pick an automation agency instead if the job is connecting tools you already use, not writing custom software. See the best AI automation agencies in Australia.
• Pick an independent consultant first if you want a recommendation before you commit to a build. See the best AI consultants in Australia.
• Pick a manufacturer-focused list if the build runs on a plant floor rather than in an office. See the best AI automation for Australian manufacturers.

Hourglass AI turns some builds down. It settled on an ideal customer segment and a repeatable system to deliver, and that meant saying no to a lot of things clients want. Many of those asks are very custom, specific AI platforms that would make a great business in themselves. If that is the build you need, one of the bigger firms on this list will serve you better. The same goes if you want everything built for you while you stay hands-off. Hourglass AI's best clients are the ones who want to learn to build with AI, and that is now an explicit filter in its sales calls. The firm started on the same bet. Co-founder Finlay Ekins skipped the grad job and rewrote the playbook from scratch to start Hourglass AI at 22, reasoning that AI is new enough for anyone willing to put in the effort to get good at it.

What to check before you hire one

Before you hire an AI development company in Australia, check four things. Does it connect to the systems you already run? Will the software take actions, or only answer questions? Where will your data live, and who owns the code? Can the firm show a measured result for a named client?

A firm that quotes a build before it has learnt how your team works is guessing. Early on, Hourglass AI kept struggling to get enough information on its clients to start building, so it switched to doing AI audits first. Now it interviews the client's team before building anything. As its February 2026 newsletter put it, everyone is seeing and hearing AI hype, but most people don't know what they don't know. Startup Daily reported Hourglass AI's offer as the opposite of the industry default: fixed scope, fixed fee and a live system in weeks.

Does it connect to the systems your business uses today?

Integration means the AI system reads from and writes to the software a business already runs, such as its CRM, accounting system, email and file storage. The business does not move onto a new platform. A firm that names the systems it connects to, and shows a client build that uses them, has done this before. Fourteen of the 40 scored companies earn full marks on engineering, which requires a named stack, a described client build and named builders. Team 400 names the Azure ecosystem, Advancer names Claude, n8n and Twilio, and Hourglass AI names HubSpot, Gmail, Slack and Xero.

The common assumption is that the big general tools will be the first software AI replaces. The builds point the other way. Salesforce, Outlook and Excel stay in use. What gets replaced is the niche software built for a particular industry, because a workflow built for one business beats a tool built for people like it. So the build you pay for should work inside the systems you keep. Ask each firm which of your systems it will read from and write to, and which ones it leaves alone.

Will the software take actions, or only answer questions?

An AI system does real work when it reads data, makes or checks a decision, and takes an action in the business's software, rather than answering questions in a chat window. The proof is a client build that names the task it took over. The Dragonfly Remedial report workflow at Synap AI, the Fortescue fleet control automation at AI Advancements and the outage approvals 4mation automated for the ABC each replace a named piece of manual work.

AI just crossed the line from demo to doing real work. That doesn't mean a firm can take a one-line brief and go off on its own. In July 2026 a Hourglass AI co-founder ran up to 8 agents at once and let them ship over 100,000 lines of code, and the supervision wore him out. Fully autonomous AI engineers that plan, write, test and ship a whole project on their own are overhyped. AI still needs context and decisions from a person to work. When a firm shows you a build, ask who on its team made the decisions, and who on yours.

Then ask how accurate the system has to be. For legal and finance builds, Hourglass AI's September 2026 update puts it bluntly: 99% right is 0% useful, and 100% accuracy is required before anyone signs off. The fix is to measure the agent and give it a test to pass. One Hourglass AI agent that kept filing information in the wrong place went from 62% right to 100% once it was measured, set a test and left to fix itself.

Where will your data live, and who owns the code?

A guardrail is a written commitment about where client data lives, who owns the code and what happens when the engagement ends. Personal information handled by an Australian business falls under the Privacy Act 1988. The Australian Government's guidance on AI adoption also lists keeping a person able to oversee an AI system and intervene as one of its essential practices. Eleven of the 40 scored companies earn full marks on ownership: they state that the client owns the code, receives handover documentation, keeps the system on its own accounts, and is not locked into ongoing support. Seven state none of the four.

Buyers often ask about hosting a model inside their own company, so their data never leaves the building. It's a fair instinct, but a Hourglass AI developer costed doing it properly at upwards of $1.6 million just to begin. The more useful questions are simpler. Whose account does the AI bill run on, yours or the firm's? And does the model provider learn from your data? When a free model turned up in August without saying who built it, the usage was the payment: hundreds of thousands of real coding sessions are exactly the data that trains the next model.

Security is the question that decides most of these projects. Hourglass AI's September 2026 update says it came up in almost every sales call, from VC funds to a 50-person industry body. The answer that moved those buyers, it says, was building on the client's own infrastructure. Ask any firm that says the same which data still goes to a hosted model such as Claude or GPT.

Can it show a measured result for a named client?

An outcome is a named client, the system that was built and the number that changed, such as hours saved per report or a share of approvals automated. Eight of the 40 scored companies earn 6 or more of 10 on evidence, and 13 score zero. A published entry price is the other proof a buyer can check before signing: Synap AI's assessment is $950, AI Advancements' scoping is $2,000 and Sandlabs' roadmap is A$1,999.

A named client with a number is the best proof a firm can publish, and it's still another company's result. The number worth having before signing is the one for your own team. That's why Hourglass AI's audit ends with a ranked AI roadmap and an ROI summary before any money goes into building.

How this list is ranked

Where engineering happens is printed as a fact on each card and never scored beyond the engineering points. EB Pearls, Enterprise Monkey, VT Digital, DianApps, Microknot, PhoenixDX, JPLoft, Saigon Technology, Jaarvis Technologies, Appther, Concetto Labs, Apptechies and Appello each say on their own sites that some or all engineering happens outside Australia.

Hourglass AI scores full marks on published prices and ownership. Without those two tests, EB Pearls would be first on 40, then 4mation and AI Advancements on 38, Mantel Group on 37, and Advancer and Team 400 on 35, with Hourglass AI on 33. If team counted double instead of engineering, EB Pearls would still be first on 58, ahead of 4mation and Advancer on 55 and Hourglass AI and Synap AI on 48.

Hourglass AI loses flatly on two tests. Its evidence score is 8, against 10 for EB Pearls and AI Advancements. On team it scores 5, against 10 for EB Pearls, 4mation, Advancer, Spark Interact, Enterprise Monkey, Microknot, PhoenixDX and Saigon Technology, 9 for Mantel Group and DianApps, and 6 to 8 for 17 others. Hourglass AI is under two years old, and its site gives a city but no street address. Its evidence score of 8 counts four named clients. Integr8 and Voyagin Australia each have a numeric result, for 3 points each. Two more are named without a number, for 1 point each: Strategy Breakdowns, for which it built a render engine, and Schweigen, which Startup Daily named as running an AI customer support system Hourglass built. It scores none of the 2 points for a Google rating. Its other published results name no client, so they score nothing under the evidence rule.

Related reading

• Where AI automation pays back for an Australian business
• Why a company AI brain is not a data warehouse
• OAIC guidance on privacy and commercially available AI products
• APP 8: sending personal information overseas
• Australia's AI Ethics Principles

References

1. Australian owned and operated since 2004 (cited for EB Pearls)
2. 320-plus engineers are full-time staff in Kathmandu, Nepal (cited for EB Pearls)
3. AI practice prices each paid stage (cited for EB Pearls)
4. AI consulting packages run $6,000 to $12,000 plus GST (cited for 4mation)
5. chatbot builds are $20,000 and $40,000 (cited for 4mation)
6. end-to-end custom agent workflow is $25,000 plus GST (cited for 4mation)
7. automated planned outage approvals (cited for 4mation)
8. QV IQ chatbot it built for Ego Pharmaceuticals (cited for 4mation)
9. guide to software IP (cited for 4mation)
10. has more than 40 staff (cited for 4mation)
11. founded in Brisbane in 2015 (cited for Advancer)
12. AI Readiness Audit starts from $2,500 (cited for Advancer)
13. Simple automation projects run $5,000 to $15,000 (cited for Advancer)
14. Voice AI agents for Bartons Motor Group handle thousands of service calls (cited for Advancer)
15. platform page (cited for Advancer)
16. Hourglass AI: We Make Australian Companies AI-Native (cited for Hourglass AI)
17. Hourglass AI: Pricing (cited for Hourglass AI)
18. Hourglass AI: Company AI Brain (cited for Hourglass AI)
19. Hourglass AI: Client Results (cited for Hourglass AI)
20. Startup Daily: Former Startmate boss Michael Batko is back in founder mode building with Hourglass AI (cited for Hourglass AI)
21. Hourglass AI News: Monthly Update, July 2026 (cited for Hourglass AI)
22. Hourglass AI News: Monthly Update, September 2026 (cited for Hourglass AI)
23. Michael Batko on Substack: I started a company (special edition, June 2026) (cited for Hourglass AI)
24. Hourglass AI News: Monthly Update, August 2026 (cited for Hourglass AI)
25. SmartCompany: Startmate CEO Michael Batko steps down after eight years (cited for Hourglass AI)
26. AI Down Under: The AI CEO Who Backed Startups Worth $3.5B+, Michael Batko (cited for Hourglass AI)
27. Finlay Ekins: Same Newsletter, New Face (cited for Hourglass AI)
28. publishes every price on its home page (cited for Synap AI)
29. built a multimodal report workflow (cited for Synap AI)
30. Code, data and hosting sit in the client's own environment (cited for Synap AI)
31. a custom field service platform with AI scheduling (cited for Team 400)
32. Custom AI projects typically cost $50,000 to $250,000 (cited for Team 400)
33. works with Azure AI Foundry, OpenAI, Anthropic Claude and LangChain (cited for Team 400)
34. AI agent FAQ (cited for Team 400)
35. Team 400: Mobile app developers, Melbourne (cited for Team 400)
36. embedded with Fortescue's fleet control team (cited for AI Advancements)
37. gets a week back every month (cited for AI Advancements)
38. begins with a $2,000 paid scoping engagement (cited for AI Advancements)
39. grown alongside its clients for two and a half years (cited for AI Advancements)
40. designed and delivered Ada, an adviser digital assistant, for Allianz Retire+ (cited for V2 AI)
41. V2 AI: Onyx Gaming, a unified data platform (cited for V2 AI)
42. V2 AI: Improving student outcomes at a top university (cited for V2 AI)
43. Consultancy.com.au: V2 AI continues to ramp up with further senior leadership recruits (cited for V2 AI)
44. commission platform that processes lender remittance from more than 24 aggregators (cited for Sandlabs)
45. starts with a A$1,999 roadmap (cited for Sandlabs)
46. AI consulting team built SteelChief's AI-powered logistics system (cited for Arcadian Digital)
47. cut its claims processing costs by 95% (cited for Mantel Group)
48. AI automation page (cited for Spark Interact)
49. AI agent development page (cited for 7 Pillars)
50. saves about 25 hours of management time a month (cited for Zenias)
51. established in Geelong in 2014 (cited for Enterprise Monkey)
52. founded in 1999 (cited for Alliance Software)
53. cut its manual data entry by 23% (cited for VT Digital)
54. built Orby, an AI agent platform (cited for DianApps)
55. Watt Realty build (cited for Elite Dev)
56. onshore and offshore teams in Sydney and India (cited for Microknot)
57. Finlay Ekins: Please Don't Take my Brain
58. Finlay Ekins: The Great AI Gap
59. Hourglass Digital AI Newsletter, 25 March 2026
60. Hourglass Digital AI Newsletter, 18 February 2026
61. Finlay Ekins: The AI Casino
62. Finlay Ekins: Humans are the Limit
63. Finlay Ekins: The Right Balance of AI
64. Privacy Act 1988
65. keeping a person able to oversee an AI system and intervene
66. Finlay Ekins: A Ferrari with No Fuel
67. Finlay Ekins: A Mysterious Ox
68. OAIC guidance on privacy and commercially available AI products
69. APP 8: sending personal information overseas
70. Australia's AI Ethics Principles