This year the common feedback was, that the sessions were really good! Maybe it was due to Sessionize, which we used for selection of the best sessions. Or maybe it was because this year we had less (even less than usual) marketing sessions or because we were a bit heavier on business and other topics. Or maybe the sessions‘ quality just went up as the speakers are getting more professionals.

During the day I didn’t see a single session, as usually, but I checked every single presentation we published on the website to get a feeling what it was about.

Are you also curious what I learnt?

Andrew Fawcett spoke about Heroku and how it makes sense to involve it in projects, especially now with the new Heroku AppLink, the new feature in Summer ’25. All the edge cases where batch/queuable classes takes too long or even triggers are slow and all that can be offloaded to Heroku to get some extra power. Good bye Functions, welcome Heroku.

Jean-Pierre Rizzi spoke about Intelligence views. Initially I’ve been surprised how such topic could make it to the agenda, because I never really understood these views and didn’t see any single useful functionality there, but JP showed me, I’ve been wrong. It is a mixture between list views and reports – the summaries, color identified key changes on records and especially the Pipeline Inspection looks pretty cool.

Matt Meyers and his communication of security risks to the C-Suite. After reading his book (and he has been selling it at CD25 and donated 5x the proceedings to the Medecins Sans Frontieres) I knew what will come, but it is still interesting to see what the C-level really cares about – reputation, lost of revenue, legal fines. And not surprisingly they have a simplified view of the situation with misunderstanding of shared responsibility model or lack of awareness of risk in Salesforce.

This year was probably first time we covered CRM Analytics and how it fits together with standard reports & dashboards in Salesforce and with Tableau. Jiri Dvorak did a good job in explaining the capabilities such as actionability of in-context actions (no need to open records first and do something later), advanced visualisations including linked filtering and much more. Without the talk it is harder to follow, but still plenty of keywords you might want to adapt and understand.

Another product we never spoke about but as several clients in the Czechia are currently running a project of its implementation it made sense to cover it. And Oleg Mastriukov did a great job in covering Loyalty Cloud Management, one of the more expensive product (45 000€/month). B2C and B2B2C use-cases, no-code design, real-time, complex, covering probably even more than typical customer needs when they think about loyalty. The Trailhead modules are also great, check them out.

Aaron Crear (btw a new Salesforce MVP) sticked with his traditional „reporting topics“ but this time extended it with Slack and how to access them. He spoke about CRM Analytics and also gave a tip how to use Flow to send reports to Slack.

Autogenerating documentation of your org – what a live saver for a lot of people. But does it really? Nicolas Vuillamy & Mariia Pyvovarchuk showed us the magical sfdx-hardis together with Agentforce, monitoring differences and admin friendly explanations of what it does. Another thing on my list of to-dos I want to play with.

Failures

Nikita Belov and his easy ways how to ruin implementation – based on the feedback one huge hilarious stand-up comedy. Nikita was also one of the lucky ones who catched our advertisement in Prague metro. Anyway, the rules are simple:

1. set deadline without knowing anything and promise the Moon
2. avoid documentation
3. go custom! And screw naming of elements (see above session from Mark) or duplicate the things
4. go directly to production, give everyone admin access,
5. and much more

Andrea Kuklová & Tomas Hnizdil didn’t stay behind and shared their 10 fuck-ups specific to discovery. Skip it completely, do it in one go, just do the same you already have just on different platform, do solutioning during discovery, take all wishes and nice-haves and pretend a lot, so client feels save.

Ankita Dhamgaya spoke about architecture anti-patterns such as automation overload (seen that been there), hammer & nail (well, …), tangled data flow with no clear rules what and why, spaghetti and much more. Looks like this year we had plenty of sessions about failures and how to achieve them quickly.

Georgy Avilov spoke about flexibility failures and when it is ok not to be agile. Project portfolio analysis, so you know where to focus, project healths assessment, clear red flags for agile (slow communication, junior team, dependencies, no clear vision and strict budget) and that you should be ready to prepare your own blend of agile/waterfall/others but it should be blended for each project.

UX Designers

Yana Abdualeva and Viktor Jamrich spoke about the new look and feel of Salesforce (read Cosmos design) and why it is important for everyone. While I agree with all the numbers – such as 88% of users will never return after a bad user experience – I still don’t get the need for Dark mode. The Figma components must be super cool for designers, but as a developer I still prefer to hardcode (yes, it is hardcode) the color into the CSS rather than learning all the new classes I can use – –slds-g-color-surface-1, –slds-g-radius-border-4. But it is a must, otherwise all the great features of branding will wanish as half of the app will not change the color 🙁

Also a big shot out to them as they prepared the new look and feel of our presentations and visul identity in general!

While the presentation title suggested that Katka Vokrinkova will speak about her maternity leave it wasn’t completely true. But there are so many similarities to UX designers and how users need to be able to predict what will happen and what they will see (hey, have buttons in the same order please!), understand what they see on screen (and you, as an UX designer need to understand them), repeat and reinforce the same patterns, and you should grow as UX designer as well.

Business Analysts

The author of several books about project discovery, project management and Salesforce consulting – Pei Mun Lim – spoke about everything and nothing and the presentation included a lot of her pictures. It was all about diagraming, process mapping, a bit of design thinking and everything else – who, what, why, when, how, empathy, clarity, connections, governance, approach, structure, blueprint, roadmap, diagrams (one for each audience), consistency and much more. Super long presentation from which you get maybe less than half (just from the PDF) but it will still make you think!

Iulian Chiriac and Alexandru Furtuna followed on the wave with mapping processes for maximum business impact. They spoke about the 80/20 rule and how only 20% of metadata in the org already deliver 80% of the value and the rest is just legacy and stuff no-one dares to delete (seen that, been there). How you need to talk with real users, not just managers. That mapping before building is crucial, but you still need to be able to iterate for the impact. That business is crucial, not which cloud you will use.

And finally Lukas Vavrin spoke about design thinking, the core principles and when it isn’t a good idea. A mindset change, when you want to solve the right problems, not what just has been asked. Involve users (a lot), define problems, ideate, protype, fail often and quickly, test! „How might we“ should not be too broad or include a solution (seen both a way too much on projects), crazy 8 (or just 4) approach where you quickly provide X variants and have something to talk about. Not ideal when the stakeholders don’t have enough capacity or in siloed organizations. All in all – would love to test it our on a project for real.

Developers

David Fernandez Rivero moved from his typical DevOps topics to triggers and flows – when to choose what? And guess what? 80 – 90% of automation needs can be covered by flow, which is now a power tool! Yes, more complex things still need to be solved with triggers, such as high-volume DML, custom sharing, complex logic and a few more most likely.

Use Flow unless there’s a clear reason to use a Trigger!

If David gave the best tips where to go with Flow and when with Trigger, Achraf El Kadiri spoke about best tips how to write Apex for scalability and maintenability. Batch, Queuable, Future, test driven development, error handling, bulkification (not surprise), service classes and much more. It is always good to repeat.

We did have some deeper developers talks as well, Mara Tyrlik was one of them – may the wire be with you. Did you know that the @wire gets updated differently based on what is the load type? How Lightning Data Service will help you? When cache expires and that you might get an updated record even before expiry?

Todd Halfpenny always has some great developer and funny talk as well with a lot of interactivity. Browser APIs for more interactivity, presentation full of recorded videos and short blocks of code so you know how.

Volodymyr Radko spoke about the new Dynamic Formula Evaluation and how it can make your code more cleaner (and less formula fields will be needed). But looks like the presentation didn’t render properly, so you need to check documentation.

Apex Cursors by Łukasz Waszkiewicz – a year old functionality which should be a game changer for large data volumes processing, nice alternative to batch classes or probably something you might solve with the magical Heroku Andrew spoke about. Just kidding.

Architects

Integrations might be an art on its own, as Prabhat Sharma correctly mentioned. Do you really know which pattern and tool to use for every scenario? Prabhat has plenty of scenarios in his presentation with explanation why and what and also including all the things you should thing about – retry, logging, authentication, monitoring and more.

One more on integration, this time Svatopluk Sejkora and view specifically on SAP and master data sync. The data model is slightly different, where the data are created is a big question including ownership of data, integration patterns and their benefits. There isn’t really one size fits all, plenty of questions can be asked.

Flows

That magical Transform element people have hard time to find the right use-cases for. Chandan Mullick luckily saved us and showed how to use it to skip loops, set values of fields for updates, aggregate data (what??!!), join collections (beware of limits but holly molly).

Barbora Sourkova presented screen flows reloaded with reactive screen actions, which is the latest addition to minimize number of clicks/screens you need to have in a flow. Well, you will have way more flows to achieve the results, but from user perspective it will be way better. Sorry, we didn’t record the live demo so you have no clue what it was about 🙁

Error handling and flows, easy, right? Markus Frohler showed some gotchas – you are sure where the transaction starts and ends, when it rolls back and such thing? I’ve been personally surprised a while back, great to hear about it from someone (as I completely missed the related Trailhead module.

Naming Conventions for Flows by Mark Jones – first of all I didn’t know that Automation Champion wrote some „Golden Standard“ back in 2021, but I can see the risk of inconsistency and trouble pin pointing down the right element just based on the debug log. There must be a reason why developers are willing to „waste“ a day on finding the right name for a variable or method, so why consultants and admins don’t worry much?

Andy Engin Utkan shared plenty of best practices for flows – before-save vs after-save, big flows, SOQL/DML in loops, loops inside loops (just check the Transform session before going further), input/output variables, system context use when not needed, fault path and rollbacks (go right to Markus‘ session), async path and when to use them and much much more.

Flow performance should be a topic as well, even though we don’t speak too often about it as admins probably don’t really think that deep. But Fabian Kramer thinks differently and gave us plenty of reasons why we should care. Live demos have been presented of Apex Log Analyser and all those FlowInterview* objects together with some general statistics – what is not that much slower on one record is way slower for mass updates. And they happen surprisingly often from my experience.

Agentforce & AI

Christian Szandor Knapp was probably the only one who slip in with Agentforce topic, but securing agent’s action was worth it. He spoke about the times back then after GA, when we tried to fix security by specific instructions not to get fooled, shared responsibility model, how the current agents are different to previous agents, how you should verify user according to your company standards (and how Summer ’25 allows you to do it). Short summary? It is heavy and completely new universe.

Well, actually Michal Verner had an AI related session as well – GitHub Copilot and plenty of recorded videos with examples, enjoy them!

Marketing

I said at the beginning that this year we had only handful of marketing related sessions, Markus Dang and his session about building a marketing roadmap. Way too often we see customers who just want to have something simple, quick, the same way as in previous system, they can do it themselves, etc. But very often they are surprised it doesn’t work. They don’t look at the journey through their customers eyes (but only internal), their data is non-existent/duplicated/just poor and they have to blast emails all over the place without clear segmentation and thinking about customer first. While omnichannel looks like an overkill for plenty of people, when your marketing effort is jumping all over the place it is just too much. Obviously a bit of AI and continues optimalization.

Melissa Shepard spoke about Data Cloud and key tips for successful implementation – it isn’t ETL or data cleansing solution (which might surprise some people), you should involve stakeholders early, start with MVP, define clear business objectives (so you know what the MVP should deliver), assess data quality and continuosly monitor it, go native (integrations) as much as possible and understand the magical topic of credits.

Melissa Hill Dees & Vicki Moritz-Henry followed with the right use-cases for Data Cloud and Agentforce. If average organization uses 991 different application, are you sure the data are connected? And when 80% of customers say experience is as important as product, do you really provide them with the best? Agentforce might not be the right answer though.

Hanna Nimchuk spoke about SFMC audit and how to get it started. Why? Better performance, stronger security, reduced costs, better customer experience. Slide 11 will give you a checklist and you can wonder – do you have journey goals set? Use splits, ideally the Einstein powered ones? Does the automation run or fail? Data extensions naming aka do you really know what is in them? Everyone has admin and you have there users who already left the company?

It was heavy!

You still here with me? That was one long article as well as one long day. We also got plenty of photos so you can follow with some doom scrolling.

Day full of learning and we hope that next year won’t be different. We already have a date – May 19, 2026 and tickets on sale, hope you will join us!

Napiš komentář, díky!