TL;DR
In this episode of Spark Sessions, Jules Love, co-founder of Spark AI, takes apart the hype around AI agents and shows what they actually do inside an agency. He sets out the six components that forms an agent, demos agents running live in Claude, ChatGPT and Gemini, and works through the parts of agency work that should stay with people.
SPARK SESSIONS EPISODE 4
Agents for Agencies
This is our fourth episode of Spark Sessions, Spark AI's monthly series. The question we hear most from agencies right now is: what does good actually look like? This series is our answer to that. Each session is grounded in a practical demo showcasing how AI is embedded into the way they work.
In this session, Jules Love, co-founder of Spark AI, works through what agents are, where they sit alongside prompts, projects and assistants, and what it takes to run them without losing control of quality. He demos agents running in Claude, ChatGPT and Gemini, so you can see the triggers, the permissions and the outputs behind each one.
Duration:
45 Minutes
Who its for:
Agency Founders, Operations Leads, and Client Services Directors
What is it about:
Agent Anatomy, Permissions and Guardrails, Workflow Design
Format:
On-Demand Video
What an agent actually is
An agent carries out a task you have defined, that can run to a schedule or a trigger. It might start every morning at seven, or the moment a status changes on a board. It reaches into your systems for what it needs, your CRM, inbox, drive or project tool. The result gets delivered to you by your preferred means, such drafted email, or a document in a folder.
Example agents for Claude, ChatGPT and Gemini
Claude:
Chief of Staff: Runs every morning, builds a brief as a web page from his email, transcripts, CRM and Monday boards.
Commitment sweep: Runs at midday, finds anything Jules agreed to and has not done, writes it into an auto captured section on Monday for review.
Competitor watch: Runs every Tuesday into a Google Doc, adding what has changed each week and what he should do about it.
ChatGPT
Deal coach: Work through his CRM, rank the next step on every deal, and leave the follow-up as a Gmail draft.
Gemini:
Four Cs research: Company, competitor, cultural and customer gems, rebuilt as a single Workspace Studio workflow.
Where the judgement stays with people
Jules's rule for handing work to AI without losing quality is a sequence. Think, AI, think. You form your own view first, of what you expect and what good would look like. AI does the heavy lifting. Then you interrogate what comes back. The judgement stays yours at both ends.
FAQs
-
Start with one someone else has built. If your team is on Google Workspace, open Workspace Studio and browse the templates. If you are in ChatGPT, the workspace agents sidebar has a set of pre-built agents you can run or copy.
-
Jules Love is co-founder of Spark AI and author of Shift: AI for Agencies, which has sold over 5,000 copies. He teaches on the leadership programme at Saïd Business School, Oxford, and has worked with more than 80 agencies in the UK and internationally on building AI into the way they operate.
-
Each session is free to attend. We announce upcoming episodes on our LinkedIn, through the Spark Intelligence newsletter, and our Events page. Recordings are available on demand via the Spark AI website after each session. If you want to make sure you don't miss one, the best way is to subscribe to the newsletter.
-
We have launched a Agent Build Sprint. It is a two-day sprint where your team learns to spot the right tasks for an agent, then builds and tests on real work. A week later, we come back to refine what you've built, map what's next, and set you up to share it with the rest of your team.
A clear path from AI experimentation to strategic advantage
At Spark AI we work with you to build internal AI capability – providing the training, the structure for business transformation and building the agents your team needs to fully integrate AI and build a permanent engine for growth.
-
Spark AI Maturity Score
Know where you stand
Your AI maturity, scored against the Spark AI Maturity Model® and benchmarked against 80+ agencies and brand teams.
Pricing
Free -
AI Foundations
Build shared foundations
From individual experimentation to team-wide capability. Hands-on AI training that raises your whole team to the same level – building shared skills, a shared language, and confidence across roles, so AI-use stops being individual and starts being organisational.Half day from
£5000+VAT -
AI Accelerator
Develop strategic capability
From scattered adoption to building the AI engine that changes how your organisation works. Ten weeks of focussed transformation and role-specific training across each of your teams to build AI into your business model.10 weeks from
£5,000+VAT per stream for team of 10
£25,000+VAT across your whole business -
Agent Build Sprint
Build working AI agents for your team
A two-day, hands-on hackathon where your team builds real AI agents for their own workflows. A team of 10 leaves with 3–4 working agents and the skills to build more.Pricing from
£10,000+VAT -
Built for you AI agents & tools
Build tools and innovate
From optimisation to innovation. Bespoke consulting and transformative custom AI tool builds – for agencies ready to scale their AI operations and supercharge growth.Pricing
Tailored to your needs -
AI Frontier
Keeping pace with change
Ongoing, high-level support for leaders as your organisation innovates, training for your AI power-users to grow and evolve capability as AI matures, and AI agents builds across your teams.Long-term partnership
From £2000 + VAT per month
