Engineers, GTM and sales specialists who ship automation and chat-based flows across WhatsApp, ChatGPT and Claude. AI-fluent by default. Financial services experienced. Available now.
All figures are live percentages from the Rafiki Talent database, updated continuously. We don't say "we have AI experts". We show you what proportion of the bench works with AI tools daily.
| Capability | % of pool | Why it matters for SaaS and fintech |
|---|---|---|
| AI-fluent (ChatGPT, Claude daily) | 70% | The single most common skill in the pool. Ahead of every design and engineering tool. |
| Engineering layer (ex. GitHub) | 44% | Shippable engineers who can build, deploy and maintain production AI systems. We support GitHub, GitLab, Bitbucket and others. |
| Design layer (Figma, 3D, holographic) | 44% | Product designers fluent in Figma, plus custom 3D and holographic design for spatial, AR and next-generation chat interfaces. |
| WhatsApp for Business | 24% | Specialists who build chat-channel distribution flows for the surface 2B+ people use daily. |
| Industry | % with experience | What this means for your brief |
|---|---|---|
| B2B | 61% | Majority of the pool has shipped for B2B buyers, sales cycles and procurement. |
| SaaS | 50% | Half the pool has built or scaled software-as-a-service products. |
| Financial Services | 43% | Fintech, payments, lending, wealth, insurance. Deep regulatory and product context. |
| eCommerce | 39% | Transactional, conversion-focused product and engineering experience. |
| AI | 33% | One in three have shipped within AI-native products or AI tooling specifically. |
| Marketplace | 25% | Two-sided product experience: supply, demand, and the friction between them. |
We don't broker engineers we couldn't replace ourselves. Every Rafiki director ships code, builds MCP servers, and lives in the same chat surfaces our clients want to deploy into.
| Capability | Tools and frameworks | Applied to |
|---|---|---|
| AI-native development | Claude Code, Cursor, Cowork | Faster shipping cycles across every client engagement. |
| Custom MCP servers | Anthropic MCP spec, TypeScript, Python | We've shipped multiple MCPs in production, including the one powering our own bench. |
| RAG and vector search | Pinecone, Supabase pgvector, embeddings | Knowledge bases that LLMs can query reliably. |
| Agentic workflows | Multi-step agents, tool use, function calling | Autonomous task execution inside client systems. |
| Workflow automation | n8n, Make, Zapier | Glue between disparate SaaS tools and AI surfaces. |
| Chat-channel builds | WhatsApp Business API, Slack, Teams | Where customers and teams actually live. |
| Browser automation | Playwright, Claude in Chrome | Agents that operate web apps on behalf of users. |
| LLM evals and observability | Langfuse, LangSmith | Production-grade AI you can debug and trust. |
Proof: our Rafiki Talent MCP server is built, shipped, and live. It's the engine behind this page's statistics, our internal sourcing workflows, and the brief-in product described below.
Petl Pay is the payment infrastructure spun out of Rafiki Works. It's live in production today with real transactions, real customers, and real rails. It's the proof point that chat-native finance isn't a roadmap claim. It's a working product.
Petl Pay handles project-based payments for contractors, agencies and subcontracted work. Subbies describe their work in chat, AI generates compliant invoices, the invoice routes to the general contractor, and payment moves over open banking rails or stablecoin corridors.
Open banking across the United Kingdom and European Union for instant bank-to-bank settlement. Stablecoin payment corridors for cross-border work where traditional rails are slow or expensive. Built on the same chat-native principles we sell to clients.
| Origin | Spun out of Rafiki Works |
|---|---|
| Status | Live in production |
| Rails | Open banking (UK/EU) + stablecoin corridors |
| Backed by | Visa Accelerator, Baobab Network, Jobtech Alliance |
Petl Pay is also the payment layer powering Rafiki's own brief-to-invoice-to-payment flow described in the next section.
Our Talent MCP is a live Model Context Protocol server that lets agencies and startups access our fractional talent pool from inside Claude, ChatGPT custom GPTs, or any MCP-compatible client. Search profiles, send briefs, and (in private beta) receive invoices and pay over open banking or stablecoin rails.
Live and operational. Access is granted by enquiry rather than via a public app store listing. Once connected to your Claude account or set up as a custom GPT, you can search the pool, view profiles, filter by skill, country, hourly rate and availability, and request shortlists for specific briefs.
Private betaBrief-sending, profile lookups, and live instant bank-to-bank payments are all in active testing. Full end-to-end automation from chat is approximately 80% complete. We're onboarding selected design partners now. Email Greg to join the private beta.
Rafiki provides on-demand fractional engineers, GTM specialists and sales talent for B2B SaaS and financial services companies. The pool is AI-fluent, with deep experience building automation and chat-based flows across WhatsApp, ChatGPT and Claude. Engagements range from a few hours per week through to 3+ days per week.
70 percent of the Rafiki talent pool uses ChatGPT and Claude as daily working tools. It's the single most common skill across the entire pool, ahead of every design and engineering tool. The management team is 100 percent proficient in Claude Code, Cursor, Cowork, building custom MCP servers, RAG, vector databases and agentic workflows.
Rafiki has gravitated toward B2B SaaS, financial services and AI. 61 percent of the talent pool serves B2B, 50 percent serves SaaS, 43 percent serves financial services and 33 percent serves AI specifically. We've also delivered work across eCommerce, marketplace, EdTech, PropTech and hospitality tech. Separate sector deep-dives are coming.
Yes. Our MCP server is live and accessible by enquiry. Once connected to your Claude account or set up as a custom GPT app, you can search the talent pool, view profiles and send briefs directly from chat. End-to-end invoicing and instant bank-to-bank payment from chat is currently in private beta.
Payments are processed through Petl Pay, our sister company, which is live in production with real transactions today. Petl Pay supports open banking rails across the UK and EU for instant bank-to-bank settlement, plus stablecoin payment corridors for cross-border work.
Rafiki talent builds across WhatsApp Business (24 percent of the pool is fluent in WhatsApp for business), ChatGPT and Claude via custom MCP servers, and Slack, Teams and Discord for internal workflow automation. The management team has shipped custom MCP servers in production, including the one powering this page.
Most engagements start with a brief sent via our enquiry form or through the MCP. We return a ranked shortlist within 48 hours, you meet the talent, and engagements begin within the week. Common engagement shapes are fractional CTO, fractional GTM lead, AI implementation specialist, MCP server build, RAG knowledge base build, and chat-channel integration projects.
Describe what you need in plain language. We return AI-fluent fractional talent matched to your sector and stack, usually within 48 hours.
