How to Choose a Claude Implementation Partner: Criteria and an RFP Template for Investment Firms
Dr. Leigh Coney
Founder, WorkWise Solutions
October 24, 2026
14 min read
TLDR: To choose a Claude implementation partner, judge firms on proof of work, send every shortlisted firm the same RFP, and start with a paid pilot that must produce a working skill on your own redacted documents in week one. There are four routes: Anthropic's own team and materials, large consulting firms with announced Anthropic alliances, specialist firms, and building in-house. Anthropic's partner tiers and certifications signal a firm's scale and training with Anthropic; they cannot show whether the people on your project understand deal data, MNPI rules or an LP DDQ. Call two references at firms like yours before you sign.
Send me the RFP in this guide and compare my answers with everyone else's. Through WorkWise Solutions I run the Claude Rollout for Private Equity and Credit Funds at a published price, $25,000 for firms of 1 to 10 people up to $40,000 for 61 to 150, over three to six weeks. When your team trains, it trains on your own redacted documents.
Table of Contents
1. The Short Answer: Proof of Work, One RFP, a Paid Pilot
Choose a Claude implementation partner on work you have seen. Shortlist from any of the four routes below, send every firm the same RFP, and start with a paid pilot that must produce a working skill on your own redacted documents in week one.
Most of a Claude rollout at an investment firm has little to do with Claude. The work is in your data rules, your templates, the settings that enforce them and training people on their own documents. The best partner understands how your firm works and can prove it fast.
Operating partners choosing help for portfolio companies can use the same method, with two extra tests in section 4. This guide ranks no firms. It gives you one page to compare them on.
A good partner is easy to spot once you ask for work instead of slides.
2. Four Routes to a Claude Rollout
Help comes from four places.
Fits when: you have an internal owner with time, your needs are standard, and you want to learn from the source.
Watch for: material written for every industry, which won't know your IC template, MNPI rules or LP questions.
Fits when: you need scale across many companies or countries, heavy systems work, or you already use one of these firms.
Watch for: who staffs your project day to day, and whether a firm your size gets the senior people from the pitch.
Fits when: you want people who already speak CIMs, IC memos and LP DDQs, a fixed scope and training on your own documents.
Watch for: a small bench. Ask who does the work, who covers when they are away, and how deep the engineering goes.
Fits when: someone at the firm owns AI, with protected hours, a partner sponsor and comfort in the admin console and GitHub.
Watch for: the day job. A rollout run in spare time competes with every live deal.
Anthropic sells Claude directly, Enterprise included, and its education team publishes free courses on Claude Academy, among them Deploying Claude Enterprise with Confidence. Ask Anthropic what hands-on help comes with your plan.
Several large firms have announced Anthropic alliances: Deloitte, Claude for more than 470,000 people (October 6, 2025); Accenture, an Accenture Anthropic Business Group with about 30,000 professionals trained on Claude (December 9, 2025); PwC, Claude Code and Cowork for its teams plus deal work for private equity sponsors (May 14, 2026); and KPMG, which Anthropic named "a preferred partner for private equity" on May 19, 2026, to build Claude-powered products together for PE portfolio companies.
Anthropic's directory lists Accenture, Bain, BCG, Capgemini, Cognizant, Deloitte, KPMG, McKinsey, PwC and others at Global Premier, its top tier, as of October 23, 2026. Ode with Anthropic, an enterprise AI services firm introduced on July 15, 2026 by Anthropic, Blackstone and Hellman & Friedman, works with mid-sized companies across sectors.
Specialists trade scale for focus; an in-house build trades fees for your people's hours. Hold all four, your own team included, to the criteria in section 4.
3. What Anthropic's Partner Directory and Certifications Tell You
Anthropic launched the Claude Partner Network on March 12, 2026. The program has four tiers: Registered, the entry level, then Select, Preferred and Global Premier. In Anthropic's words, "Partnership begins at Select," and the Claude Partner Network Directory lists partners from Select up.
Each tier is defined by counts, set out in Anthropic's June 3, 2026 announcement:
- Select needs "at least 10 active certified individuals, at least 2 joint customers deployed in production in the trailing 12 months, and at least 1 public customer story."
- Preferred needs 100 certified people, 15 deployed customers and 3 public stories.
- Global Premier needs 1,000 certified people, 100 deployed customers across three or more regions, 15 stories and a joint business plan.
A certified person counts only with a current certificate and Claude use in the past 90 days. There are four exams: Associate, Foundations ($99), "For consultants, sellers, and delivery leads who guide customers toward the right Claude use cases"; Developer, Foundations ($125); Architect, Foundations ($125); and Architect, Professional ($175). Certificates last 12 months, and Anthropic's certification page adds: "Certification is currently available only to organizations in the Claude Partner Network."
Read all of it as signals. A tier shows a firm's certified capacity and production work with Anthropic, and Global Premier's 1,000 certified people is a number only large firms can reach. A certificate shows a person passed Anthropic's exam in the last year. Neither shows whether the people on your project know how a deal team handles MNPI, what LPs ask in a DDQ or how your IC wants a memo laid out.
Certification is open only to network members, so even a strong in-house owner cannot hold one. Ask who on your project holds which exam, and count a current certificate in a firm's favor.
Use the directory to build a longlist. Use the RFP and the week-one skill to choose.
4. The Criteria That Matter for an Investment Firm
Seven criteria separate a partner who can roll out Claude at an investment firm from one who can roll it out anywhere:
- Deal data and MNPI rules. Can they write a one-page data rule your partners will sign, then set the settings that back it up?
- Plan choice. Enterprise adds SCIM, custom roles, audit logs, the Compliance API, custom retention, IP allowlisting and per-user spend limits on top of Team. A good partner tells you which you need, in writing.
- Connectors to your data room and document stores. An owner enables each connector, each person authenticates individually, and tool permissions can be Always allow, Needs approval or Blocked. Ask where they would start. Read-only is the answer you want.
- Skills on your templates. Ask what they would change in Anthropic's private equity plugin, and listen for your IC template and screening criteria (the Claude skills guide covers this).
- Governance documents an LP DDQ asks for. A policy, the data rules, a settings record and training records your IR team can paste into a questionnaire.
- Training on your own documents. Sessions on redacted CIMs, models and memos, by role, teach more than any tour of features.
- Fluency in private markets. Screening, QoE, the IC process, capital calls and, for credit funds, covenants, without a glossary.
Operating partners hiring for portfolio companies should add two tests. Can the partner run one program across five to fifteen companies with local examples? And does each company keep its own Claude organization and data?
Few of these criteria test Claude trivia. Most test whether the partner understands how your firm works, which is harder to learn.
5. Proof of Work: A Working Skill in Week One
Slides are easy to make and hard to check. A working skill is the opposite, which makes it the best test you have.
Make the first deliverable of any pilot a skill that runs on your own redacted documents by the end of week one. Two good candidates:
- A screening skill that scores three redacted CIMs against your written criteria, in your screening note format.
- An IC memo skill that drafts a past deal's memo in your house template, with every figure tied to a source document.
Provide redacted documents under NDA, your criteria or template and a pilot group, then judge the output with the people who would use it:
- Does it follow your template, or a generic one?
- Does every number point to a source, and did it flag what it could not find?
- Did the partner set connectors to read-only and explain each setting it changed?
- Did they tell you what the skill gets wrong before you found it?
Pay for the pilot. A paid week is a real project with a real scope, the only way to see how a firm works. Andrew Ng's advice, quoted on this page, points the same way: start with pilots that win early.
A partner who cannot produce one working skill in a week will struggle to produce twenty in six.
6. The RFP Template: 25 Questions
Send every shortlisted firm these questions with one deadline, ask for written answers, and compare them group by group.
The team
- Who will do the work, by name and role, and how many hours a week will each give us?
- Which of them have worked at a PE firm, a credit fund or a portfolio company, and in what role?
- Which of them hold a current Anthropic certification, which exam, and when does it expire?
- Who covers when our main contact is away, and how fast do you answer questions?
Claude depth
- Which plan, Team or Enterprise, would you recommend for us, and which features decide it?
- Show us a skill you wrote. What does its SKILL.md say, and how did you test it?
- What would you change in Anthropic's private equity plugin for us, and how would you distribute our copy?
- How would you set connector tool permissions and the Cowork approval settings for our deal teams?
- Anthropic announced on September 16, 2026 that chat and Cowork are merging. What changes for us, and when?
Data and security
- How will you write our deal-data and MNPI rules, and which settings will back them up?
- Which retention settings do you recommend, and what happens to a chat after someone deletes it?
- Will your staff see our data during the project, and under what agreement?
- What do you advise about local Cowork history on laptops, which admins cannot centrally delete?
- Which governance documents will we have at the end, ready for an LP DDQ?
Delivery and timeline
- What is the week-by-week plan, and who does each step?
- Will you build a working skill on our redacted documents in week one?
- What do you need from us: people, hours, documents and admin access?
- At handover, who owns the settings, the skills and the repository they live in?
Training and adoption
- Is training run on our own redacted documents or on generic examples?
- How does training differ for partners, deal teams, IR and finance?
- How will we measure adoption at day 30, and from which data?
- What support do we get in the weeks after launch?
Commercials
- Is the price fixed for the scope, and what would change it?
- Do you resell licenses or take referral fees from any vendor?
- Who owns the skills, templates and code you build for us?
A firm that answers all 25 in plain words, with names and dates, has already shown you how it will communicate for the next six weeks.
7. Red Flags
Five answers should slow you down:
- License resale with referral fees. A partner paid more when you buy more seats or a bigger plan has a reason to recommend both. Question 24 asks directly.
- Promises that Claude keeps no copy of your data. Deleted chats leave Anthropic's back end within 30 days, Enterprise retention can be set no shorter than 30 days, a thumbs up or down rating stores the whole conversation for up to 5 years, and local Cowork history sits on the laptop. A partner who gets this wrong will write it into your DDQ answers.
- Demo-only. Polished demos on sample data and reluctance to touch your redacted documents.
- No fixed price for a defined scope. Published prices make comparison easy; at minimum, get a fixed price in writing before workshops start.
- Nothing after launch. No office hours, no 30-day measure and no handover of skills and settings, in the weeks when adoption is won or lost.
Add a sixth: any promise that a setup makes your firm compliant. Settings and policies help, and your CCO and counsel still own the conclusions.
Each flag has the same root: a partner whose incentives or knowledge stop at the sale.
8. Reference Checks
Ask each finalist for two references at firms like yours, and call them yourself. Confidentiality is normal in private markets, so if names are off limits, ask for a call with a client's COO under NDA, or for redacted work such as a policy or a skill.
On the call, ask:
- What was working at day 30, and how did you know?
- Who did the work: the people from the pitch, or others?
- What broke, and how did the partner handle it?
- Did the governance documents hold up in an LP DDQ?
- Would you hire them again for the same job?
Select-tier partners need at least one public customer story, so read the ones on offer and look for a firm like yours.
End each call with one question: what would you do differently? The answer tells you what the partner did not.
9. What It Should Cost
Keep two budgets apart: what you pay Anthropic and what you pay a partner.
Anthropic publishes its seat prices on its pricing page. A Team standard seat costs $25 a month, or $20 billed annually. Enterprise is $20 a seat a month billed annually plus usage, with a 20-seat minimum when bought self-serve, and the Help Center is blunt: "All usage across your organization is billed together at API rates, regardless of who consumed it."
For a small firm on Team, licenses are the small line: ten standard seats billed annually cost 10 x $20 x 12 = $2,400 a year. Setup, training and your own people's hours decide the total.
For partner fees, one public benchmark: WorkWise Solutions, which publishes this guide, lists every price on its pricing page. Its Claude rollout costs $25,000 for firms of 1 to 10 people, $30,000 for 11 to 25, $35,000 for 26 to 60 and $40,000 for 61 to 150, over three to six weeks. Treat it as one reference point among the quotes you collect.
Ask every finalist to price the same scope, with exclusions such as licenses and data subscriptions listed, and compare only then; the rollout cost guide builds full year-one budgets. Prices for different scopes describe different projects.
10. Running the Selection in Three Weeks
Three weeks is enough if you prepare before you send anything.
- Before week one: a one-page brief with firm size, plan status, three workflows, your data rules and a decision date.
- Week one: a longlist across the four routes, your own team as the benchmark, and the RFP out to three to five firms with a five-business-day deadline.
- Week two: answers compared, a shortlist of two, and a walkthrough of a skill each has built.
- Week three: reference calls, then a paid pilot with a working skill due at the end of its first week.
Name one person to run it, usually the COO or an operating partner, and bring one deal partner to the walkthroughs. The people who will use the skills should judge them.
You learn more from a week of real work than from a quarter of meetings.
11. Where to Start
- Write the one-page brief, with your three workflows and your data rules.
- Pick the routes that fit, using the figure in section 2.
- Send the 25 questions to two or three firms, and answer them for your in-house option.
- Make a working skill on your documents the week-one deliverable of a paid pilot.
- Call references and compare prices on the same scope.
- Sign a first phase with a 30-day measure and a clean exit.
The rollout playbook covers what happens after you sign, and the AI consultant guide covers help beyond Claude.
WorkWise Solutions, which publishes this guide, is a specialist in private capital, the third route above, with 30+ engagements behind it. It publishes every price, sells no licenses and takes no referral fees. Its Claude Rollout for Private Equity and Credit Funds costs $25,000 for firms of 1 to 10 people up to $40,000 for 61 to 150 (by proposal above that), over three to six weeks, and its training runs live online on the firm's own redacted documents.
Not a fit if you want a partner that also resells licenses or staffs a large engineering team on site.
"Execute pilot projects to gain momentum. Rather than starting with a massive, multiyear project, it is more important to get the AI flywheel spinning with early successes."
Andrew Ng, AI Transformation Playbook (Landing AI, read October 2026)
- •The best test of a Claude implementation partner is a working skill built on your own redacted documents in week one of a paid pilot, judged by the people who will use it.
- •There are four routes to a Claude rollout: Anthropic's own team and materials, large consulting firms with announced Anthropic alliances, specialist firms and building in-house, and many firms combine them.
- •Anthropic's partner tiers are defined by counts of certified people, deployed customers and public stories, so they signal a firm's scale and training with Anthropic rather than its fit for an investment firm.
- •Anthropic's four certification exams cost $99 to $175, last 12 months and are currently available only to organizations in the Claude Partner Network, so a certificate is one signal among several.
- •Send every shortlisted firm the same RFP, so the answers can be compared line by line, and ask each finalist to price the same scope.
- •Red flags include license resale with referral fees, promises that Claude keeps no copy of your data, demo-only pitches, no fixed price for a defined scope and no plan for the weeks after launch.
- •Claude licenses are rarely the large line: ten Team standard seats billed annually cost $2,400 a year, so setup, training and adoption decide what a rollout costs.
Frequently Asked Questions
How do I find a Claude implementation partner?
Look across four routes: Anthropic's own team and free materials, large consulting firms with announced Anthropic alliances, specialist firms that know private markets, and your own people. Anthropic's Claude Partner Network Directory lists partners by tier, from Select up to Global Premier, and peers at other funds can name firms they used. Send two or three shortlisted firms the same RFP, ask each finalist to walk you through a skill it has built, and call references at firms like yours. Then start with a paid pilot whose first deliverable, due in week one, is a working skill on your redacted documents.
What should a Claude rollout include for a private equity firm?
At minimum: a written plan decision between Team and Enterprise, deal-data and MNPI rules on one page, retention and chat-rating settings, connectors to the data room and document stores set read-only first, a few skills written around your own templates such as the IC memo, the governance documents an LP DDQ asks for, and training on your own redacted documents. It should end with a named internal owner, a 30-day measure and support after launch. The rollout playbook covers the sequence.
Do Claude consultants need Anthropic certification?
No. Certification is a useful signal and nothing more. Anthropic offers four exams, from Associate, Foundations at $99 to Architect, Professional at $175, and says certification is currently available only to organizations in the Claude Partner Network, so an in-house team cannot hold it. Each certificate lasts 12 months. A current certificate shows a person passed Anthropic's exam; it says nothing about deal data, MNPI rules or an LP DDQ. Ask which people on your project hold which exam, then judge the firm on a working skill built on your documents.
Related Guides & Articles
Claude Rollout for Private Equity and Credit Funds
Team or Enterprise set up with deal-data and MNPI rules, connectors, a starter skills library and LP DDQ governance documents. From $25,000, three to six weeks.
What a Claude Rollout Costs an Investment Firm
Licenses, usage, setup and training for 10, 25 and 60-person firms, with the seat arithmetic shown.
Claude Skills for Private Equity
What Anthropic's PE plugin does command by command, and how to fork it around your own templates.
How to Choose an AI Consultant for Private Equity
Seven firm types compared on fit, cost, timeline and limits, with a scorecard and red flags.
How to Roll Out Claude Across Your Firm
A 90-day plan: data rules first, one beachhead workflow, champions, training and measurement.
AI Readiness Assessment Providers for Private Equity
What an assessment should deliver, who provides one, typical cost and timeline, and red flags.
Want one more set of answers to compare?
Send me your RFP, or the 25 questions in this guide, and I'll answer every one in writing. The Claude Rollout for Private Equity and Credit Funds sets up Claude Team or Enterprise with your deal-data and MNPI rules, retention settings, connectors to your data room and document stores, a starter skills library and the governance documentation an LP DDQ asks for. It costs $25,000 for firms of 1 to 10 people, $30,000 for 11 to 25, $35,000 for 26 to 60 and $40,000 for 61 to 150, over three to six weeks, and any training I run uses your own redacted documents.