How to choose between a consultancy, an outsourcer, software and Meo
A healthcare provider that wants AI agents working in its operations has at least six options besides Meo: a large consultancy, a marketing agency, an offshore revenue cycle or outsourcing firm, an AI software product, its own team, or the AI features its EHR vendor ships. Each is the better choice in some situations. This article sets out what has changed in the market, what to ask each option, and when we think each one wins, Meo included.
Embedded engineers are now common
The large AI companies now sell the approach Meo uses. OpenAI launched a deployment company in 2026 that starts with a focused diagnostic, picks a few priority workflows, then puts forward-deployed engineers inside the client to build them (OpenAI, 2026). Anthropic launched an enterprise AI services firm with Blackstone, Hellman & Friedman and Goldman Sachs the same year (GIC, 2026). Tomasz Tunguz, citing LinkedIn data, reports that forward-deployed engineer job postings rose about 42 times from 2023 to 2025 (Tunguz, 2026).
So the label tells you little. What separates one embedded team from another is the mechanics: who does the work, how success is counted, what runs where, and what happens when you want to stop.
Outcome-linked fees are common too
Tying fees to results is now mainstream in consulting. In November 2025 McKinsey's managing partner for the UK, Ireland and Israel said about a quarter of the firm's global fees come from outcome-based pricing (Business Insider via Yahoo Finance, 2025). When any firm offers outcome pricing, read the clause. It should name the metric, the measurement window, the data source, the threshold and how disputes are settled. How Meo's fees work is on our pricing page.
The alternatives, fairly
A large consultancy
Large firms bring brand, depth, mature legal and risk functions, and indemnity. They often start with a strategy or diagnostic phase, with the build as a later phase or run with a technology partner. Choose one when you need an enterprise-wide change program and a big name behind it.
A marketing agency
Agencies are strong on creative, media buying and campaigns, usually on a monthly retainer plus media spend. They rarely build the systems behind marketing, such as intake that answers every inquiry or attribution that counts new patients in your CRM. Choose one when your bottleneck is demand and creative. Meo can work alongside your agency.
An offshore revenue cycle or outsourcing firm
These firms now sell AI agents together with human staff, at large scale. In KLAS's 2025 review of end-to-end revenue cycle outsourcing, clients of some vendors reported struggling to hit expected cash-collection targets and inconsistent quality from offshore teams, and clients had yet to see significant operational or financial impact from vendors' AI investments. The same review says these engagements typically involve rebadging staff and last at least five years (HIT Consultant, 2025).
Offshore handling of patient data brings its own rules. Medicare Advantage plans must collect attestations for offshore subcontractors that handle beneficiary data, and some states bar or restrict offshore work on Medicaid data (Shumaker, 2025). Choose an outsourcer when you want to hand over an entire function, people included.
An AI software product
For one well-defined, high-volume workflow, a product is often the fastest route. One patient-access voice vendor states a go-live of 30 to 45 days (Assort Health blog, 2026), faster than our typical 8 to 16 weeks for a first workflow. Products also often publish security attestations. Choose one when a mature product matches your workflow. Meo fits better when the work crosses functions or depends on systems no product connects to.
Building it yourself
You own everything and know your systems best. The costs are hiring, retention and permanent maintenance. MIT NANDA's 2025 report found that internal builds succeeded about one-third as often as tools bought from specialized vendors or built with partners, a self-reported, correlational finding (Fortune, 2025). HFMA calls bot maintenance the biggest factor organizations overlook in automation projects, because bots must be retrained whenever the systems they work on change (HFMA, 2019). Choose this when AI is your product and you can hire and keep the team.
Your EHR vendor's AI
At least one major EHR vendor now ships its own agents and has previewed a platform for building more inside the EHR (Fierce Healthcare, 2026). Use them where the work lives inside the EHR and the feature already exists. Much of intake and billing happens outside it, in phone lines, faxes, payer portals, the CRM and marketing. We build around the EHR features you already use.
The critique of embedded engineers
Critics have a point. Some operators call forward-deployed engineering "consulting with a new label," and N47 argues the model only works when field work turns into reusable product (N47, 2025). Others worry about lock-in: once engineers build on one AI vendor's interfaces, switching can mean rebuilding (Tunguz, 2026).
Our answers are mechanics you can check. The engineer who maps your process builds it, with a design group from your team that meets weekly. Success is a dated baseline and targets your owners accept before the build. Agents can run in your environment and connect to the systems you keep. Each phase ends with your sign-off, and any line can switch back to today's process. Ask for the ownership and exit terms in your proposal. The how we work page shows every step.
When to choose each
- A large consultancy, when you need an enterprise-wide change program with a large brand behind it.
- A marketing agency, when demand and creative are the bottleneck.
- An outsourcing firm, when you want to hand over a whole function, people included.
- An AI software product, when one high-volume workflow matches a mature product.
- Your own team, when AI is your product, and you can hire and keep the team.
- Your EHR vendor's AI, when the work lives entirely inside the EHR.
- Meo, when several workflows need to change inside the systems you already run, with your people deciding and results measured against your own baseline.