The Three Pricing Categories That Actually Exist
Most buyers walking into AI consulting conversations expect to hear one answer to "how much does this cost." There is no one answer. AI consulting in 2026 exists in three pricing categories that differ by an order of magnitude or more.
**Traditional enterprise consulting** runs $200 to $500 per hour for senior consultants, with engagements typically in the $50K to $2M+ range. Firms in this category include McKinsey QuantumBlack, BCG X, IBM Consulting, Accenture, Deloitte, and smaller boutique enterprise AI shops. Engagements are project-based, scope is negotiated upfront, and delivery is measured by milestone completion. The buyer is typically a Fortune 500 or large-enterprise organization.
**Managed AI services** run $0 to $5,000+ per month on subscription pricing. Providers in this category (Sentie, Single Brain, and a growing field of similar firms) ship pre-built AI agents that integrate with the client's existing tools, with ongoing management bundled into the subscription. The buyer is typically an SMB or mid-market business. There are no separate implementation invoices; everything is in the monthly fee.
**DIY with off-the-shelf tools** runs $0 to $200 per month for tool subscriptions (ChatGPT Plus, Claude, various AI writing or automation tools). The buyer self-implements, self-maintains, and self-troubleshoots. There is no consultant involved. The financial cost is small; the time cost is significant.
Understanding which category you are buying in is more important than negotiating within a category. A buyer who walks into a McKinsey conversation expecting managed-service pricing will be confused; the same buyer walking into a Sentie conversation expecting enterprise-grade custom model development will misread what they are getting. Get the category right first.
Traditional Enterprise Consulting: $200 to $500 per Hour, $50K to $20M Engagements
Traditional AI consulting is structured around custom work. The consulting firm sends a team (typically 4 to 10 people) to study your business for several weeks or months, design a custom AI implementation, build it, integrate it with your systems, and hand it off to your team to operate.
Hourly rates vary by firm tier. Tier-one firms (McKinsey QuantumBlack, BCG X) bill senior consultants at $500 to $1,500 per hour with team-blended rates around $400 to $600. Tier-two firms (Accenture, Deloitte, IBM) bill at $250 to $500 per hour with blended rates around $200 to $350. Boutique AI shops bill at $200 to $400 per hour but often with smaller teams.
Engagement structure determines total cost. A typical breakdown:
- **Discovery and strategy** (4 to 8 weeks): $40K to $300K. Output is an AI strategy deck and a recommended implementation roadmap. - **Implementation phase one** (3 to 6 months): $200K to $2M. Output is one or two production AI systems deployed. - **Ongoing optimization** (annual retainer): $200K to $1M+ per year. Output is continued support and additional model development.
For a Fortune 500 company running an enterprise-wide AI transformation, total annual spend with a tier-one firm runs $5M to $20M+ in years one and two. For a mid-sized public company doing a more contained AI transformation, $1M to $5M is the typical range.
When traditional enterprise consulting is the right fit: you are a Fortune 500 or large enterprise, you have complex custom model development needs that off-the-shelf foundation models cannot serve, you need significant organizational change management, you have an existing data science team to maintain what gets built, and you have a budget envelope above $1M for the work.
When it is the wrong fit: anything below that scale.
Managed AI Services: $0 to $5,000 per Month
Custom AI agent builders took the enterprise consulting model and inverted it. Instead of custom model development priced as a project, managed services pre-build common AI agents and price them as a subscription. The economics work because the agents are templated and the operational overhead is amortized across a customer base, not a single engagement.
Pricing tiers in this category are converging across providers:
- **Free tier**: $0 per month. Typically includes assessment, account setup, and limited sandbox usage. Sentie's free tier specifically includes full business analysis and Business Brain setup, with the limitation being that you cannot run production agents against live data until you upgrade. - **Starter tier**: $199 to $499 per month. Includes 1 to 3 production agents, basic integrations, and minimal Success Manager time. Sentie's Starter is $499 per month with unlimited production agents and unlimited integrations (capped on usage credits). - **Pro tier**: $499 to $1,500 per month. Includes expanded agent capacity, full integration library, dedicated Success Manager, and priority support. Sentie's Pro is $899 per month. - **Custom/Enterprise tier**: $2,000 to $10,000+ per month. Includes unlimited agents, multiple Success Managers, custom integrations, white-glove onboarding, and SLAs. Sentie's Enterprise pricing is custom.
What is bundled into the monthly fee: implementation, integration, ongoing optimization, monitoring, agent updates, and human Success Manager time. There are no separate implementation invoices and no surprise integration fees. The provider takes the operational risk; you pay a predictable monthly cost.
When custom AI agent builders are the right fit: you are an SMB or mid-market business ($1M to $500M revenue), you have repeatable operational processes that need automation, you do not have or want to build an internal AI team, you want a predictable monthly cost rather than a project-based commitment, and you want to start within weeks rather than months.
When it is the wrong fit: you need custom machine learning model development on your proprietary data (Sentie and similar providers configure foundation models; they do not train new models from scratch), you have unique regulatory or air-gapped infrastructure requirements, or you have an enterprise-scale change management problem.
DIY with Off-the-Shelf Tools: $0 to $200 per Month
The DIY path is real and underrated. For a single operator or very small team, the right answer is sometimes "sign up for ChatGPT Plus and figure it out yourself" rather than hiring any kind of consulting.
The financial cost is minimal. ChatGPT Plus is $20 per month. Claude Pro is $20 per month. A meaningful tool stack for solo automation rarely exceeds $200 per month total.
The time cost is significant. You are reading documentation, watching YouTube tutorials, experimenting with prompts, troubleshooting integrations, and accepting that your AI work will be amateur-grade for the first 6 to 12 months while you learn. For an experienced operator who enjoys this work, that learning curve is acceptable. For someone whose time is better spent on their core business, the time cost is the real expense.
When DIY is the right fit: you are a solo operator or 1 to 3 person team, you have the technical comfort and interest to learn AI tooling yourself, your operational volume is below the threshold where managed consulting makes sense, your AI use cases are simple (drafting emails, generating ideas, summarizing meetings), and you do not have integrations that need to run unattended in production.
When it is the wrong fit: you have AI use cases that require integration with multiple business systems running unattended (CRM, helpdesk, scheduling, etc.), you have volume that exceeds what manual chat-style usage can handle, you are not the right person to be doing AI configuration work given your other priorities, or you need consistency and reliability that hand-rolled tool chains rarely provide.
Many small businesses start DIY and graduate to managed services around the time their volume or complexity exceeds what one person can babysit. That is a reasonable progression.
What Drives Cost Variance Within Each Category
Knowing the category gets you to a rough price range. Within each category, several factors push you toward the higher or lower end of that range.
**For traditional enterprise consulting:** - Firm tier: Tier-one (McKinsey, BCG X) commands 2x to 3x the rates of tier-two firms - Scope: data science work is 3x to 5x more expensive than configuration work - Industry: heavily regulated industries (financial services, healthcare, pharma) command premium rates - Customization: highly custom implementations cost more than reference-architecture implementations - Geography: US/UK rates are higher than offshore equivalents but with quality and IP differences
**For custom AI agent builders:** - Number of agents in production (more agents = higher tier) - Volume of agent activity (usage credits in plans like Sentie's) - Integration count (more integrations = more Success Manager time) - Custom playbooks/skills (custom configuration may bump to Pro or Enterprise tier) - SLA requirements (priority support and 1-hour response push to Pro+) - Multi-team deployment (multiple Success Managers = Enterprise)
**For DIY:** - Number of tools in stack (each adds monthly subscription cost) - API call volume (if hitting API rate limits, may need higher tiers) - Time cost of self-learning and self-maintenance (the largest cost most operators underestimate)
For most small to mid-market buyers, the right question is not "how can I negotiate this down" but "am I in the right pricing category to begin with." Buyers attempting to negotiate enterprise consulting down to a $50K project usually end up disappointed; buyers attempting to scale custom AI agent builders to enterprise complexity often find the model breaks. Pick the category that matches your situation and the price range follows.
The Hidden Cost: What Cheap AI Actually Costs You
The cheapest option in any category is not always the lowest total cost. There are real hidden costs that buyers often miss when comparing AI consulting options.
**Hidden cost of DIY when your operations need more**: A solo operator who tries to run customer support through DIY tools when their volume is actually too high will lose customer satisfaction, miss inbound opportunities, and burn out trying to babysit the tools. The financial cost of DIY is low; the cost of inadequate execution can be significant.
**Hidden cost of cheap custom AI agent builders**: Some custom AI agent builders offer cheaper plans with severely limited agent capacity, no Success Manager time, and minimal integration support. The monthly price looks attractive but you end up with agents that do not perform because no one is tuning them. This is why Sentie's free tier is genuinely free (no time limit, full setup support) and the Starter tier includes unlimited production agents rather than artificially capping them.
**Hidden cost of consulting firms that under-deliver**: Traditional consulting engagements that produce a 50-page strategy deck and no working systems represent significant sunk cost. The financial cost was paid; the implementation cost is still ahead. Watch for consulting firms whose primary deliverable is a slide deck.
**Hidden cost of switching providers later**: If you start with one provider and need to switch, you pay onboarding costs twice and lose the institutional context the first provider had built up. This matters more in custom AI agent builders where the provider accumulates business knowledge in their Business Brain or equivalent.
The honest framing: cheap is fine when it matches your actual needs, but cheap is expensive when it does not. A $200 per month managed service that delivers reliable AI agents is better economic value than a $20 per month tool stack that does not, and that is better value than a $50K consulting engagement that produces a deck. Match the spend to the actual operational need.