Tech consulting is at an inflexion point: AI is driving a shift from billable hours to reusable patterns, productised offerings, and outcome‑based models. Agentic workflows, where AI agents can plan, decide, and act across tools, are accelerating this shift away from the traditional bums-on-seats style engagements.
By the start of 2025, successful projects had already moved AI from open secret to embedded tooling. These projects treated agents as strategic professional tools focused on specific delivery tasks.
As the year progressed, forward-thinking projects began evolving their AI use beyond tactical deployment and made huge strides toward genuine Agentic workflows. For many, this begged the question, “Do we even need consultancies anymore?”
This piece examines how Service-as-Software is emerging as a disruptive trend in the consulting industry. It provides practical recommendations on how consultancies can adapt to this existential crisis without sacrificing their core values or profit margins.
What Are Agentic Workflows?
AI agents are seen by many as the next evolution of SaaS. Agentic workflows are AI-driven processes where autonomous AI agents plan, decide, and act through a series of steps to achieve a goal with minimal human hand-holding.
Agents are different to chatbots like ChatGPT. In an agentic workflow, you don’t just call an LLM once; you give one or more agents a goal, tools, context and constraints, and they dynamically choose which steps to run next based on context and intermediate results. Instead of a fixed linear script, the workflow can branch, loop, or revise its own plan as it works toward an outcome.
The temptation is to treat this as “more automation.” But agentic workflows are something different; well designed agents can perceive, decide, and act across systems. They quietly reshape who does the work, how value is measured, and what clients expect to buy from you.
How Agentic Workflows Impact Consulting
For years, IT consultancies were hired for two things:
- Effort: The grunt work: running processes, writing scripts, maintaining runbooks, working the queues. Essentially expanding resource capacity for clients.
- Expertise: shaping strategy and decisions, so clients could focus on their core business.
That translated into a simple operating model: specialist projects plus outsourced BAU, delivered through a mix of junior and senior people on time‑and‑materials contracts.
Agentic workflows cut straight through the first part of that equation.
Agents are good at:
- Grinding through repetitive analysis and verification
- Orchestrating routine flows that used to live in runbooks
- Surfacing problems early, with more context than a wall of logs
- Frontier models excel at reasoning and high-stakes task execution
So a big chunk of classic consulting activities gets commoditised: manual triage, boilerplate reporting, repetitive regression‑style passes. The grunt work that used to justify long engagements is exactly the work agents excel at.
What becomes more valuable is the work that sits above the agents:
- Designing robust pipelines where agents can safely operate
- Defining policies, guardrails, and escalation paths
- Choosing where autonomy makes sense
- Recognising the high-stakes mission-critical decisions that agents mustn’t take
Customers don’t want effort days anymore. They want a future‑proof, agent‑ready operating model, human judgment, and someone to take responsibility for how it behaves when the stakes are high.
Who Feels This Most
The firms most exposed to this shift aren't the small specialists. They're the large generalist consultancies whose core competitive advantage has always been scale: the ability to put hundreds of people on a problem across geographies.
Agentic workflows democratize that scale. A boutique firm with sharp domain expertise and well-designed agent playbooks can now deliver the throughput that used to require a 50-person delivery team. The moat that justified the Big 5 premium is narrowing, not because they're doing anything wrong, but because the thing that made them hard to compete with is becoming a commodity.
From Billable Hours to Outcome-Based, Productized Services
If agents can handle a chunk of the day‑to‑day work, billing by the hour starts to make less sense for everyone. You can’t credibly talk about efficiency and then quietly hope nothing gets too efficient. Well, I say you can’t, but many still do.
I’ve seen novel commercial patterns starting to replace the old “time and materials” mindset:
Outcome‑based engagements
Success is tied to lead time, failure rates, or stability, not how many stories were worked on. The engagements that stick are the ones with a direct line to something the client's CEO or CFO actually care about: revenue growth, cost reduction, or customer satisfaction. Everything else is activity.
Agent playbooks as products
Reusable playbooks such as autonomous incident triage, continuous readiness checks, and self‑healing health checks that you can implement across clients with predictable results. These look and feel like software products, even if they’re delivered as part of a consulting engagement.
Managed agent services
Consultancies own the configuration, monitoring & observability, and safe rollout of agentic workflows, and customers pay for reliability rather than raw hours
The shift from selling time to selling outcomes and agent‑powered service patterns is one of the core ideas for the next wave of IT consulting and presents a fundamental industry shift.
Firms that experiment with these delivery models now are already having very different conversations with their customers, and are much harder to dislodge when the next “AI platform” comes along.
Foundations First. Agents Second.
There’s a catch: agentic workflows amplify whatever you plug them into. If your systems, processes, or telemetry are a mess, agents will ship risky changes faster and bury teams in noise.
The teams that saw real value in 2025 all did the same boring things first:
- Cleaned up delivery pipelines and processes so changes were predictable
- Invested in trustworthy controls, tests and data
- Simplified monitoring so alerts meant something
Only then did they start weaving agents into specific, painful parts of the lifecycle: slow feedback loops, noisy incidents, and manual handoffs between tools and teams.
Treat “agentic” as the final stage of a maturity curve, not the first step. This lets you start small, measure impact, and make deliberate choices about where more autonomy actually makes sense.
Human Consultants Are Still Vital
As agents take on more of the mechanical work, the role of human experts shifts. We become orchestrators rather than throughput producers, choosing the high-level workflows agents should run, what “good” and “safe” looks like, and when to intervene.
That includes:
- Reviewing and curating agent actions around production, security, and data
- Tracking outcomes instead of celebrating how many tasks an agent completed
- Evolving roles: agent orchestrators, prompt engineers, decision playbook designers, AI‑aware engineers, operators and testers
This human layer is where consulting firms can have an outsized impact.
Real value lies in helping clients redesign teams and roles, and implementing feedback loops so agents can augment expertise rather than quietly work around it.
Where Aster Fits
Agentic workflows aren’t hype. They are becoming the operating model for modern delivery teams.
At Aster, our stance is simple:
- Fix the fundamentals so agents operate on solid ground.
- Help customers identify and target real pain points. Do not chase "AI everywhere."
- We apply 20 years of expertise to enforce control: use deterministic systems where outcome certainty is non-negotiable.
- Apply software engineering first principles to build reliable, scalable, and safe autonomous agents.
- Keep human judgment central, especially when decisions touch customer data, revenue, or risk.
With this approach, we’re helping customers move from people‑heavy, T&M‑driven engagements to agent‑ready operating models, outcome‑based contracts, and productised agent patterns they can roll out across their portfolios.
If you’re already experimenting with agents or working out where to start, we’d love to compare notes. Together we can chart a path that’s powerful, but not reckless.