AI Consultancies & GTM Firms: What Changed (Oct 2026)
Oct 6, 2026 · Chad Thomas
Bottom line
Money in AI services is moving from advice to implementation, from hourly fees to outcome pricing, and from big junior teams to small senior ones. Five shifts define the market in late 2026:
- SMBs try AI widely but rarely embed it. Only 14% of Goldman's 10KSB alumni have AI in core operations, and the Census puts real adoption near 20%.
- Implementation beats strategy. Less than 20% of McKinsey's work is now pure strategy, and forward-deployed engineer (FDE) job postings rose 1,165% in 2025.
- Pricing is moving to outcomes. About 25% of McKinsey's fees are outcome-based; Omnicom is replacing hourly fees with output pricing.
- Model vendors now sell services. OpenAI ($4B) and Anthropic ($1.5B) each launched deployment ventures in May 2026, aimed at the mid-market.
- GTM work is becoming engineering. Clay is valued at $7.1B and “GTM engineer” is a new job title, while AI SDRs have a mixed record.
For Blackline AI, the gap is small firms that are below what the new vendor ventures target. They need hands-on builds, training and fixed-scope offers priced on results.
What SMBs want from AI now
Most small businesses have tried AI, but few pay for it or have built it into how they work. That gap is the market for a consultancy.
The surveys measure different things. Census and JPMorgan count firms that actually use or pay for AI; owner surveys count anyone who “uses” a chatbot. Only 12% of US businesses paid for dedicated AI tools in 2021–2025, but 86% of 2024 payers kept paying in 2025.
- •Use cases: marketing, admin and customer service lead (Intuit). Among staff at 2–250-person firms, 62% use AI for customer service and marketing, 54% for operations and 51% for finance (Business.com).
- •Budgets are small: median AI spend is about $30 a month (JPMorgan). Thryv finds 53% spend $100+ a month and a third raised budgets this year.
- •Barriers are skills and trust, not cost: privacy and security, fear of errors, and not knowing what AI can do (Intuit). Goldman adds lack of expertise and difficulty choosing tools.
- •What they ask for: 73% want more training and implementation help (Goldman); 70% need more training, and 57% learn from YouTube or social media (Thryv).
- •Size matters: adoption is rising at firms with 20+ staff (32–37% at 100+ employees) but flat below 20 (Census).
How AI consultancies are changing
Clients now pay for working systems, not decks. Firms are reorganizing around implementation and outcome pricing, and model labs have entered the services market directly.
| Firm | What changed | Source |
|---|---|---|
| McKinsey | ~25% of fees outcome-based; <20% pure strategy; 25,000 AI agents beside ~40,000 staff | Business Insider, Nov 2025 |
| BCG | AI and tech services are 40%+ of $14.4B revenue, growing 25% | BCG, Apr 2026 |
| Bain | AI-enabled work ~30% of revenue, target 50% | TheStreet, May 2026 |
| Accenture | $865M restructuring; managed services growing 6% vs consulting 3% (FY26) | Fortune, Sep 2025 |
| OpenAI | $4B Deployment Company, bought Tomoro (~150 FDEs) | SiliconANGLE, May 2026 |
| Anthropic | $1.5B services venture with Blackstone, H&F, Goldman for mid-sized firms; $100M Claude Partner Network | CNBC, May 2026 |
- •Pilots that fail drive demand. MIT NANDA found 95% of firms saw no P&L return from GenAI, though its method is disputed. Bought tools succeed about twice as often as internal builds (67% vs 33%). Menlo Ventures reports 76% of AI use cases are now bought, not built.
- •Agent projects face a shakeout. Gartner expects 40%+ of agentic AI projects to be cancelled by 2027 and calls most vendors “agent washing.”
- •Clients are pushing back on fees. UniCredit cut consultant spend 24% in H1 2026; Accenture shares are down 27% this year (Irish Times/FT, Aug 2026).
- •Errors carry a cost. Deloitte refunded A$440K to Australia's government over AI-generated mistakes in a report.
How GTM firms and agencies are changing
Go-to-market work is shifting from headcount to tooling. Agencies are shrinking while a new “GTM engineer” role builds automated pipelines.
- •GTM engineering is a real category. Clay raised $115M at a $7.1B valuation in Sep 2026, more than double its Aug 2025 price, with 17,000+ customers. It reports about 100 GTM-engineer job listings a month.
- •AI SDRs have not replaced reps. 11x listed customers it did not have, ZoomInfo said its AI SDR did “significantly worse” than human reps, and the CEO stepped down in May 2025. Most positive AI SDR stats come from vendor blogs.
- •Agencies are cutting staff. WPP has cut nearly 11,000 roles since early 2025. Omnicom-IPG plans to go from about 120,000 to 105,000 staff by end of 2026 and is moving to output-based pricing.
- •Agency fees are shrinking. Gartner puts agency fees at 19.2% of marketing spend, falling to 15% by 2030, and says the staffing pyramid is “inverting.”
- •Entry-level roles are thinning. UK Big Four graduate intakes fell 6–33%. Stanford finds a 13% relative employment drop for 22–25-year-olds in AI-exposed jobs.
Implications for Blackline AI
The big firms and the new lab ventures are going after the mid-market and PE-backed firms. Small businesses with under 250 staff are mostly left with YouTube. That is where Blackline AI fits best.
- Sell implementation, not strategy. Lead with fixed-scope builds such as a customer-service agent, a lead-intake workflow, or back-office automation. MIT found back-office work gives the best ROI.
- Price on results. Offer a pilot at a fixed fee, then a monthly retainer tied to a measured metric such as hours saved or tickets resolved.
- Bundle training. 70–73% of SMB owners say they need it, and it builds trust on privacy and accuracy.
- Configure proven tools before building custom ones. Bought tools succeed about twice as often, so custom code should be the exception.
- Offer GTM engineering as a service line. Build Clay-style enrichment and outbound pipelines with a human reviewing them, rather than reselling AI SDRs.
- Target firms with 20–249 staff first. That is where measured adoption is rising, and budgets go beyond $30 a month.
- Use the research papers as proof. They show that Blackline can engineer real systems, which sets it apart from prompt-only agencies.
Sources
- • Census Bureau, AI use by businesses (May 2026)
- • Federal Reserve FEDS Note, Monitoring AI adoption (Apr 2026)
- • JPMorgan Chase Institute, AI use by small businesses (Apr 2026)
- • Intuit QuickBooks, 2026 AI Impact Report (May 2026)
- • Goldman Sachs 10,000 Small Businesses survey (Mar 2026)
- • Thryv SMB AI survey (Jun 2026)
- • US Chamber, Empowering Small Business (Aug 2025)
- • NFIB tech and AI report (Jul 2025)
- • SBE Council, AI agents in small business (Sep 2026)
- • Business.com SMB AI workplace study (Jan 2026)
- • MIT NANDA GenAI Divide coverage (Aug 2025)
- • Menlo Ventures, State of GenAI in the Enterprise (Dec 2025)
- • Gartner agentic AI forecast coverage
- • TechCrunch, Anthropic and OpenAI services ventures (May 2026)
- • Channel Insider, Claude Partner Network (Mar 2026)
- • Bloomberry, 1,000 FDE job postings (Nov 2025)
- • Business Insider via Yahoo, McKinsey CEO on AI (Jan 2026)
- • Accenture Q4 FY26 earnings release (Oct 2026)
- • LeadDev, Deloitte rolls out Claude (Oct 2025)
- • The Drum, WPP job cuts (Sep 2026)
- • BestMediaInfo, Omnicom workforce plan (Sep 2026)
- • City AM, Big Four graduate cuts (Jun 2025)
- • Stanford SIEPR, Canaries in the Coal Mine (Aug 2025)
- • Technori, Clay Series D (Sep 2026)
- • TechCrunch, Clay on GTM engineers (Oct 2026)
- • TechCrunch, 11x customer claims (Mar 2025)