AI development partner for a large company

This page is for a large company that wants to add AI to the systems it already runs: its customer records, its order and accounting software, its data stores and the older systems that hold its core records. That work differs from building a new product, because the AI has to read data from systems that were built years ago, and every change has to pass the company's own security reviews and approvals.

The firms below were matched to this situation by a sentence on their own pages about existing or older systems. Reveneau is first and Fluxon second. The other seven follow in the order of the research notes, which gives no ranking. Each page was opened on 21 October 2026. For a longer list of firms that serve large companies in one region, read AI development companies for enterprises in Silicon Valley and large consultancies with an AI practice.

What a large company needs from a partner

Four things matter more here than in the other situations in this guide.

1. Access to old systems. A large company may run software that was written decades ago. A mainframe, a large central computer that runs core records, is one example. The partner has to be able to read data from these systems without breaking them.
2. Approvals. Large companies have security teams, change boards and supplier reviews. A partner has to fit those steps into its plan.
3. Size and continuity. A large project may need many people for a long time, and the buyer has to know who stays when people leave.
4. Clear records. A public company's filings give its legal name, its headquarters and its staff count, which a buyer can check.

One survey describes how companies use outside firms. Deloitte's 2024 Global Outsourcing Survey reports on "more than 500 executives globally" [23]. It says 83 percent of the executives surveyed use AI as part of the services they buy from outside providers. The survey page read for this guide does not say what size of company they work for. Deloitte sells outsourcing advice, so the figure describes Deloitte's own sample and comes from a firm with an interest in the result.

The firms at a glance

Firm	What its pages say about existing systems	Offices, and who says so	Staff, and who says so	Anthropic tier
Reveneau	An enterprise service	San Francisco and Vancouver (own site)	Not stated	Not listed (searched 20 October 2026)
Fluxon	"Improve core systems without replacing everything at once"	San Francisco (Stripe listing, address from Fluxon)	200 (Forbes, as of March 2026); "300+ builders" (own careers page)	Not listed
Capgemini	AI to "analyze, refactor, and even rewrite legacy systems"	Paris, registered seat (French company register)	417,600 at 30 June 2026 (own press release)	Global Premier
West Monroe	Data platforms that "integrate with existing systems"	Chicago and eight other US cities (own offices page)	Not found in a primary source	Not listed
Grid Dynamics	"modernizing legacy systems" among its consulting work	San Ramon, California (own footer); Zug listed as "Headquarter" (own contact page)	4,961 at 31 December 2025 (results filed with the SEC)	Select
Bounteous	Teams that build on "existing systems and workflows"	Frisco, Texas, and Boston (own US page)	"5,000-employee" and "more than 5,000" (own pages)	Preferred
Thoughtworks	Titles on moving "legacy systems" and mainframes	San Francisco office (own office page, read 20 October 2026)	"10,000+" (own about page)	Preferred
Endava	Delivery "rebuilt" around AI, by its own account	London, registered office (Form 20-F)	11,479 at 30 June 2025 (Form 20-F)	Not listed
Cognizant	A section on AI and older systems	Teaneck, New Jersey, headquarters (Form 10-K)	351,600 at 31 December 2025, stated as approximate (Form 10-K)	Global Premier

A Form 10-K or 20-F is the annual report a listed company files with the US Securities and Exchange Commission. Anthropic's directory lists the service firms in Anthropic's partner program, and Anthropic sets each tier. It held 321 entries when read on 21 October 2026 [1].

The firms

1. Reveneau

Reveneau works from two offices: 435 Utah Street in San Francisco, and Floor 33, 909 Mainland Street, in Vancouver. It has a service page for enterprises and one for AI development. Reveneau says it writes 100 percent of its code with AI. It says each change must pass a large eval suite, a set of automated tests that Reveneau writes from the specification, before it is released. It says the suite is graded by Jev, and that Jev makes the suite run ten times faster than its previous grader. Reveneau says it uses AI instead of hiring more engineers and puts that saving into its clients' prices. It says it takes responsibility for the whole project, through production, the live system staff and customers use, and after release. Reveneau has no entry in Anthropic's directory: all 321 entries were searched on 20 October 2026 [1].

For a large company, ask Reveneau to show how the rules of your existing systems, such as which records may change and who must approve a change, would be written into the specification, since Reveneau says its tests are written from it. Ask which of your systems it would change, and how its tests would run against them before release.

Fits: a large company that wants AI work written by AI and tested against a written specification, from a firm that says it takes responsibility for the project through production and after release. You can contact Reveneau to describe the systems involved.

Pick someone else if: your supplier rules require a firm listed in Anthropic's directory, or offices in Europe or Asia.

2. Fluxon

Fluxon's enterprise page says, "We help enterprise teams move important technology forward, from new products and AI to the systems your business already depends on." [2] The same page says, "Improve core systems without replacing everything at once." The same enterprise page shows case studies for Google, Zapier and Agilon Health, and no outside source read for this guide confirms those clients. Fluxon's listing in Stripe's partner directory, written by Fluxon, gives an address at 236 West Portal Ave in San Francisco [5]. Forbes's company page gives 200 employees, as of March 2026 [3], and Fluxon's careers page says "300+ builders" [4]. Fluxon has no entry in Anthropic's directory as read on 21 October 2026 [1].

Fits: a large company that wants its core systems improved one part at a time, alongside new AI products.

Pick someone else if: you need a supplier with thousands of staff, or one listed in Anthropic's directory.

3. Capgemini

Capgemini is listed in the French company register as Capgemini SE, a European company, with its registered seat at 11 Rue de Tilsitt in Paris [7]. A Capgemini press release with figures at 30 June 2026 says, "At June 30, 2026, the Group’s total headcount stood at 417,600" [6]. That figure is the company's own, from a press release. Its data and AI page says it can apply AI "to analyze, refactor, and even rewrite legacy systems like mainframes and custom ERP code" [5]. Legacy systems are old systems a company still runs. ERP software runs a company's orders, stock and accounts. To refactor code means to restructure it so it is easier to change, while it keeps doing the same job. Anthropic's directory places Capgemini at the Global Premier tier [1].

Fits: a company with mainframes or custom ERP code that it wants analysed and rewritten with the help of AI.

Pick someone else if: you want a small team in which you meet everyone who works on the project. Capgemini's own count is 417,600 people.

4. West Monroe

West Monroe lists nine US offices on its offices page, among them 311 W Monroe St, 14th Floor, in Chicago, and offices in New York, Boston, Seattle, Los Angeles, McLean in Virginia, Dallas, Minneapolis and San Francisco [9]. Its AI services page offers Intellio Flow, which it describes as a way to build "scalable, secure data platforms that integrate with existing systems" [8]. The same page says AI-led redesign "cuts maintenance by" 30 to 50 percent and "speeds time-to-market". That figure is West Monroe's own, and the page gives no basis for it. No staff count from a primary source was opened for this guide. West Monroe has no entry in Anthropic's directory as read on 21 October 2026 [1].

Fits: a US company that wants a data platform built that connects to its existing systems, from a firm with offices in nine US cities.

Pick someone else if: you need offices outside the United States, or a partner listed in Anthropic's directory.

5. Grid Dynamics

Grid Dynamics says on its homepage that it is publicly traded on the Nasdaq stock exchange. Its homepage footer gives its headquarters as San Ramon, California, while its contact page lists an office in Zug, Switzerland, as "Headquarter", along with engineering centres in Poland, the UK, the Netherlands and other European countries. In a results release it filed with the US Securities and Exchange Commission, it says, "Total headcount was 4,961 employees as of December 31, 2025", and reports full-year 2025 revenue of $411.8 million [11]. Its homepage says, "Grid Dynamics partners with leading enterprises to deliver measurable business outcomes through AI, data, cloud, and digital transformation." [10] The same homepage says Beijing Teamsun Technology, a Chinese company, is a minority shareholder that owns 20.0 percent of the shares through a subsidiary, a figure the page marks as approximate. Anthropic's directory places Grid Dynamics at the Select tier [1].

Fits: a large company that wants AI, data and cloud work from a firm that reports its results through US filings.

Pick someone else if: your supplier rules limit firms with a foreign shareholder of that size. Ask Grid Dynamics for its current shareholders.

6. Bounteous

Bounteous names offices at 6303 Cowboys Way, Suite 375, in Frisco, Texas, and 10 High Street, Suite 705, in Boston on its US page [13]. Its Anthropic partner page calls the firm "a 5,000-employee business that runs on Claude" [14], and its directory text says "more than 5,000 team members worldwide" [1]. Both counts are its own. Its page on what it calls agentic business reinvention opens with this problem: "Many organizations are experimenting with AI but struggle to move beyond scattered tools and isolated productivity gains." [12] It describes teams that include forward deployed engineers, meaning engineers who work inside the client's company, and that build on the client's existing systems and workflows. Anthropic's directory places Bounteous at the Preferred tier [1].

Fits: a large company that runs many small AI trials and wants them built into its existing systems.

Pick someone else if: you want one small project with a small team. Its agentic page is written for organisations running many AI trials.

7. Thoughtworks

Thoughtworks gives its San Francisco office as The Mills Building, 220 Montgomery Street, Suite 1000 [18]. Its office, about and press pages cited here were read on 20 October 2026. Its about page counts "10,000+ talented Thoughtworkers around the globe" and "47 offices in 18 countries" [16]. Its press release of 13 November 2024 says a transaction took Thoughtworks private, and values the deal at $1.75 billion, a figure the release marks as approximate. The buyers, it says, were affiliates of investment funds advised by Apax Partners [17]. Its homepage, read on 21 October 2026, lists an article titled "How Thoughtworks combines AI/works™ with AWS Transform to modernize legacy systems for the AI era" and a resource titled "Accelerating mainframe modernization with GenAI" [15]. Neither was opened for this guide. AWS is Amazon's cloud service, and GenAI means generative AI, the kind that writes text or code. Anthropic's directory places Thoughtworks at the Preferred tier [1].

Fits: a large company that wants old systems, mainframes among them, moved to newer ones with the help of AI, by a firm with offices in many countries.

Pick someone else if: you want a supplier whose accounts are published through a stock exchange. By its own press release, Thoughtworks is privately owned.

8. Endava

Endava is Endava plc, registered in England and Wales under number 5722669, with its registered office at 125 Old Broad Street in London, according to its annual report on Form 20-F for the year to 30 June 2025 [20]. That filing gives 11,479 employees, directors included, at 30 June 2025, with 1,082 delivery employees in Cluj-Napoca and 909 in Bucharest, both in Romania. It is the latest annual filing found, so those figures were more than 15 months old when read. Endava's page on its Dava Flow method says the firm "rebuilt delivery around" AI [19], which is its own description of how it works. Endava has no entry in Anthropic's directory as read on 21 October 2026 [1].

Fits: a large company that wants delivery teams in the UK and Romania from a firm that files annual reports with the SEC.

Pick someone else if: you need current-year staff figures. The latest annual filing read here covers the year to 30 June 2025.

9. Cognizant

Cognizant has its worldwide headquarters in Teaneck, New Jersey, according to its annual report on Form 10-K for 2025 [22]. The same filing gives a headcount of 351,600 at 31 December 2025, stated by Cognizant as an approximate figure, with 256,900 in India, 41,600 in North America and 14,600 in Continental Europe. Its Enterprise AI Builder page has a section on AI modernization, which says older systems hold the information that AI tools need and that the tools cannot reach it [21]. Anthropic's directory places Cognizant at the Global Premier tier [1].

Fits: a large company that wants AI added to older systems by a supplier with staff in North America, Europe and India.

Pick someone else if: you want a small firm. Cognizant's own filing counts 351,600 staff.

Questions about your existing systems

Send every firm the same questions, in writing, with one date for answers:

• Which of our systems will you read from, and which will you change?
• How will you get access, and who in our company approves it?
• How will each change be tested against our existing systems before release?
• How will your plan fit our security review, our change board and our supplier review?
• Which company will sign, where is it registered, and where can we read its latest filing?
• Who will work on the project, from which country, and who replaces a person who leaves?
• Who owns the code, the tests and any data you prepare?

The AI partner situation checklist has these questions in a form to print. If the work also falls under rules set by a regulator, read the regulated industry page too. For the general method of choosing, read which type of AI development firm to hire.