Salesforce-native firms are quoting fixed bids
SiliconANGLE reported from Dreamforce on September 21 that Salesforce-native professional services firms are moving from hourly billing toward fixed-bid pricing. Cheryl Knight's reporting on Certinia at the show records the vendor going from 10 packaged agents and 64 Veda Intelligent Actions in April to 24 agents and 135 actions. Certinia is an independent software vendor building on the Salesforce platform, and none of this came from Salesforce.
Two people should read that differently. If you bill your time, a fixed bid cuts the link between your hours and your firm's revenue. If you buy consulting time, it decides who absorbs the cost when a project runs three weeks longer than anyone planned.
The reported theme is that labour demand per project is falling while project volume rises. Certinia's growing library is the supply-side evidence for the first half. The second half rests on what the delivery firms say they see in their own pipelines.
The headcount claim that runs the wrong way
Mike Piehl, chief executive of Platinum River Innovations LLC, gave SiliconANGLE the piece's most interesting claim, because it points away from the obvious conclusion. Less labour per project normally means a firm needs fewer people.
"What we're seeing in practice is that the labor demand on an individual project is coming down, which normally would mean you need fewer people. However, because the overall price points are coming down as well and the value delivery is going up, it actually means that we have more client demand as a result. So, more projects coming in, and the net effect is we need more people."
That deserves to be taken seriously rather than waved off as optimism from a man with a firm to sell. It also deserves an accurate label. One executive is describing his own business, which differs from market data, and the piece offers no headcount series across Salesforce-native firms.
Elasticity is the load-bearing assumption
The mechanism has a specific requirement. Lower prices must bring in enough extra projects to more than replace the hours each project no longer needs. Demand has to be genuinely elastic, meaning a backlog of work companies wanted done and could not justify at the old price.
Plenty of enterprise backlogs look that way. Every platform owner keeps a list of changes that never cleared a business case, and a lower quote moves some above the line. It stops working where the buyer holds a fixed annual budget and spends less of it, or where the limit was never price but the customer's capacity to absorb releases.
A consultant deciding whether to believe this should watch their own booking pipeline rather than the conference stage, since that is the honest measurement for anyone whose income depends on how platform consulting gets bought. If quoted prices fall and signed project counts stay flat for two quarters, the elastic half of the argument is failing in that market.
A fixed bid moves delivery risk onto the supplier
For a buyer, fixed-bid pricing sounds unambiguously good. The number in the contract is the number you pay, and overruns stop being your problem. Both are true, and neither makes the deal cheaper, because a supplier carrying delivery risk prices that risk into the bid.
The second effect matters more. A firm on a fixed bid defends scope far harder than one billing by the hour, because every unbudgeted change comes out of its own margin. The rate card stops being what you negotiate over. The contract's definition of done decides whether your March request counts as included work or a change order.
So requirements discipline before signature now buys more than any concession on price. Running a discovery workshop that surfaces real requirements and writing the results into a statement of work someone can be held to beats shaving a few percent off the quote. The same discipline decides whether pricing a unit of agentic work means anything once invoices start.
Packaged action counts measure supply
The counts give the pace some shape. Ten packaged agents and 64 Veda Intelligent Actions in April, then 24 agents and 135 actions by Dreamforce in September. Both libraries roughly doubled inside five months, a real engineering cadence.
Certinia also plans more than 300 actions by the end of the year. That figure is a roadmap commitment rather than shipped capability, and a delivery plan should treat it that way. Vendor plans slip, get resequenced, or arrive in a shape one customer cannot use.
A count of packaged actions measures supply. It says how much a vendor has built and nothing about how many customers switched one on or kept it running past the pilot. Adoption figures would be the interesting number, and the piece carries none, so 135 proves nothing about production.
Context engineering is the expertise that replaces the hours
If agents absorb the configuration hours, something has to remain that a client still pays for. Tyler Philpot, vice president of solution architecture at Thunder, named it in the piece.
"Context engineering has become the discipline that matters, since precise context yields better agent answers while overly broad context invites hallucinations." That describes real work. Deciding which records, fields, documents and policies an agent can see, and which it must never reach, takes someone who understands both the data model and the business process.
That expertise does not resemble traditional configuration on a timesheet, a problem for firms pricing by deliverable count and for buyers judging whether a bid is fair. Grounding and context design for enterprise agents sits closer to data architecture than to clicking through setup screens, and the people who do it well are not cheap.
Ask any services firm quoting a fixed bid to name, in the statement of work, which agents and which packaged actions they plan to use, and what happens to the price if a planned action has not shipped when their team reaches that phase. A firm that can answer has priced its own risk. One that cannot has priced yours into a number nobody has tested.
